Professional Documents
Culture Documents
Exact Solutions in Multidimensional Grav PDF
Exact Solutions in Multidimensional Grav PDF
Exact Solutions in Multidimensional Grav PDF
Vitaly N. Melnikov†
Abstract
Multidimensional gravitational models containing certain combinations of scalar fields, an-
tisymmetric forms and multicomponent perfect fluid are considered. The manifold has the
form M = M0 × M1 × . . . × Mn , where Mi are Einstein (e.g. Ricci-flat) spaces (i ≥ 1). The
exact solutions in the model are reviewed (e.g. cosmological, spherically-symmetric, black
holes and branes etc) and billiard representation near the singular point is considered. The
consideration is based on the sigma-model approach. Observational windows of p-branes and
extra dimensions are discussed.
Preface
In this book we continue our systematic studies of multidimensional models of gravitation and
cosmology mainly with fields of forms (p-branes), based on exact solutions. Previous results of
investigation of gravitational and cosmological models without or with matter sources in the form
of perfect and viscous fluids for arbitrary number of extra dimensions see in [1,2]. Some results
obtained for this latter sources, obtained after 1995 are also presented in this book. Besides new
exact solution and their investigation special attention is paid here to the observational problems
of p-branes and extra dimensions. Other results obtained within this program will be published
elsewhere. I am grateful to my colleagues V.D.Ivashchuk, K.A. Bronnikov, V.R. Gavrilov and
M.A. Grebeniuk with whom most of the results were obtained, and Prof. M. Novello and Prof.Dr.
H.Dehnen for their hospitality.
Contents
1 Introduction 7
1.1 Problem of Stability of G . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.1.1 Absolute G measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.1.2 Data on temporal variations of G . . . . . . . . . . . . . . . . . . . . . . . . 8
1.1.3 Nonnewtonian interactions (EP and ISL tests) . . . . . . . . . . . . . . . . . 8
1.2 Multidimensional Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3 Time variations of G in different theoretical models . . . . . . . . . . . . . . . . . . 11
1.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.3.2 Ġ in (4 + N )-dimensional cosmology with multicomponent anisotropic fluid 13
1.3.3 Scalar-tensor cosmology and variations of G . . . . . . . . . . . . . . . . . . 16
1.3.4 Estimation of satellite orbit dependence on the time variations of G on the
level Ġ/G = 10−13 per year . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.3.5 Compensation of non-gravity accelerations . . . . . . . . . . . . . . . . . . . 20
1.3.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2 The model 23
2.1 The action and equations of motion . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.2 Ansatz for composite p-branes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.3 The sigma model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.3.1 Restrictions on p-brane configurations. . . . . . . . . . . . . . . . . . . . . . 25
2.3.2 Sigma-model action for harmonic gauge . . . . . . . . . . . . . . . . . . . . . 26
2.3.3 Sigma-model with constraints. . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.3.4 General conformal gauges and d0 = 2 case. . . . . . . . . . . . . . . . . . . . 30
12 Appendix 263
12.1 Appendix 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
12.1.1 Ricci-tensor components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
12.1.2 Riemann tensor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
12.1.3 Riemann tensor squared (Kretchmann scalar). . . . . . . . . . . . . . . . . . 264
12.1.4 The cosmological case. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
12.1.5 Parameter C = C(b). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
12.1.6 Conformal transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
12.2 Appendix 2. Product of forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
12.3 Appendix 3. Simple finite dimensional Lie algebras . . . . . . . . . . . . . . . . . . 267
12.4 Appendix 4: Solutions for Toda-like system . . . . . . . . . . . . . . . . . . . . . . 269
12.4.1 General solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
12.4.2 Solutions with block-orthogonal set of vectors . . . . . . . . . . . . . . . . . 270
12.5 Appendix 5: Solutions with Bessel functions . . . . . . . . . . . . . . . . . . . . . . 271
12.6 Appendix 6. Killing equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
CBPF-MO-003/02 6
Bibliography 280
CBPF-MO-003/02 7
1 Introduction
The necessity of studying multidimensional models of gravitation and cosmology [1, 2, 3] is moti-
vated by several reasons. First, the main trend of modern physics is the unification of all known
fundamental physical interactions: electromagnetic, weak, strong and gravitational ones. During
last decades there was a significant progress in unifying weak and electromagnetic interactions, some
more modest achievements in GUT, supersymmetric, string and superstring theories.
Now theories with membranes, p-branes and more vague M- and F-theories [6, 7, 8, 9, 10, 11]
are being created and studied. Having no any definite successful theory of unification now, it is
desirable to study the common features of these theories and their applications to solving basic
problems of modern gravity and cosmology.
Second, multidimensional gravitational models, as well as scalar-tensor theories of gravity, are
the theoretical framework for describing possible temporal and range variations of fundamental
physical constants [4]. These ideas originated from earlier papers of Miln (1935) and P.Dirac (1937)
on relations between fenomena of micro and macro worlds and up till now they are under a thorough
study both theoretically and experimentally.
Bearing in mind that multidimensional gravitational models are certain generalizations of general
relativity which is tested reliably for weak fields and partially for strong fields (binary pulsars) up
to 0.001 (they may be viewed as some effective scalar-tensor theories in simple variants in four
dimensions [4]) it is quite natural to inquire about their possible observational or experimental
windows [12]-[19]. From what we already know, among these windows are :
– possible deviations from the Newton and Coulomb laws,
– possible variations of the effective gravitational constant with a time rate less than the Hubble
one,
– possible existence of monopole modes in gravitational waves,
– different behaviour of strong field objects, such as multidimensional black holes, wormholes
and p-branes,
– post-Newtonian parameters, standard cosmological tests etc.
As no accepted unified model exists, in our approach we adopt simple, but general from the
point of view of number of dimensions, models based on multidimensional Einstein equations with
or without sources of different nature:
– cosmological constant,
– perfect and viscous fluids,
– scalar and electromagnetic fields,
– plus their interactions,
– fields of antisymmetric forms (related to p-branes) etc.
Our main objective was and is to obtain exact solutions (integrable models) for these model
self-consistent systems and then to analyze them in cosmological, spherically and axially symmetric
cases. In our view this is a natural and most reliable way to study highly nonlinear systems. It is
done mainly within the Riemannian geometry. Some simple models in integrable Weyle geometry
and with torsion were studied too [2].
Even at that time other existing on 100ppm level measurements deviated ¿from this value more
than their uncertainties [14]. During last years the situation, after very precise measurements of
G in Germany, New Zealand and other countries, became much more vague. Their results deviate
from the official CODATA value at minimal and maximal values. This means that either the limit
of terrestrial accuracies is reached or we have some new physics entering the measurement procedure
[15, 16]. First means that we should shift to space experiments to measure G [17] and second means
that more thorough study of theories generalizing Einstein’s general relativity is necessary.
connected with overall M-theory or even some F-theory. In all these theories 4-dimensional gravita-
tional models with extra fields were obtained from some multidimensional model by a dimensional
reduction based on the decomposition of the manifold
M = M 4 × Mint ,
where M 4 is a 4-dimensional manifold and Mint is some internal manifold (mostly considered as a
compact one).
The earlier papers on multidimensional cosmology dealt with multidimensional Einstein equa-
tions and with a block-diagonal cosmological metric defined on the manifold M = R × M0 × ... × Mn
of the form n
X
g = −dt ⊗ dt + a2r (t)g r
r=0
r
where (Mr , g ) are Einstein spaces, r = 0, . . . , n [33]–[65].
In [43, 44, 48, 53, 54, 60, 72, 73] the models with higher dimensional ”perfect-fluid” were con-
sidered. In these models pressures (for any component) are proportional to the density
µ ¶
ur
pr = 1 − ρ,
dr
r = 0, . . . , n, where dr is a dimension of Mr . Such models are reduced to pseudo-Euclidean Toda-like
systems with the Lagrangian
Xm
1 k i
L = Gij ẋi ẋj − Ak eui x
2 k=1
and the zero-energy constraint E = 0. In a classical case exact solutions with Ricci-flat (Mr , g r ) for
1-component case were considered by many authors (see, for example, [41, 42, 53, 54, 72, 73, 103] and
references therein). For the two component perfect-fluid there were solutions with two curvatures,
i.e. n = 2, when (d1 , d2 ) = (2, 8), (3, 6), (5, 5) [114] and corresponding non-singular solutions from
[118]. Among the solutions [114] there exists a special class of Milne-type solutions. Recently some
interesting extensions of 2-component solutions were obtained in [115].
It should be noted that the pseudo-Euclidean Toda-like systems are not well-studied yet. There
exists a special class of equations of state that gives rise to the Euclidean Toda models. First such
solution was considered in [73] for the Lie algebra A2 = sl(3). Recently the case of An = sl(n+1) Lie
algebras was considered and the cosmological solutions were expressed in terms of new representation
[113, ?], solutions with other Lie algebra see in [333].
The cosmological-type solutions may have regimes with: i) spontaneous and dynamical compact-
ifications; ii) Kasner-like and billiard behavior near the singularity; iii) inflation and izotropization
for large times (see, for example, [112, 72]).
Near the singularity one can have an oscillating behavior like in the well-known mixmaster
(Bianchi-IX) model. Multidimensional generalizations of this model were considered by many au-
thors (see, for example, [32, 79, 80, 84]). In [85, 86, 87] the billiard representation for multidimen-
sional cosmological models near the singularity was considered and the criterion for the volume of
the billiard to be finite was established in terms of illumination of the unit sphere by point-like
sources. For perfect-fluid this was considered in detail in [87]. Some topics related to general
(non-homogeneous) situation were considered in [88].
Multidimensional cosmological models have a generalization to the case when the bulk and shear
viscosity of the ”fluid” is taken into account [116]. Some classes of exact solutions were obtained,
in particular nonsingular cosmological solutions.
CBPF-MO-003/02 10
ĤΨ = 0,
where Ψ is the so-called ”wave function of the universe”. The conformally-covariant form of this
equation for the vacuum comology with n spaces of constant curvature was treated first in [55] (for
conformally-covariant WDW equation see also [67]) and integrated in a very special situation of
2-spaces.
The WDW equation for the cosmological constant and for the ”perfect-fluid” was investigated
in [71, 112] and [72] respectively.
Exact solutions in 1-component perfect fluid case (with scalar field) were considered in detail
in [103]. In [63] the multidimensional quantum wormholes were suggested, i.e. solutions with a
special-type behavior of the wave function (see [68]).
These solutions were generalized to account of cosmological constant in [71, 112] and to the
perfect-fluid case in [72, 103]. In [87] the ”quantum billiard” was obtained for multidimensional
WDW solutions near the singularity. It should be also noted that the ”third-quantized” multidi-
mensional cosmological models were considered in several papers [62, 111, 103]. One may point out
that in all cases when we had classical cosmological solutions in many dimensions, the corresponding
quantum cosmological solutions were found also.
Cosmological solutions are closely related to solutions with the spherical symmetry. Moreover,
the scheme of obtaining them is very similar to the cosmological approach. The first multidimen-
sional generalization of such type was considered by D. Kramer [93] and rediscovered by A.I. Legkii
[94], D.J. Gross and M.J. Perry [95] ( and also by Davidson and Owen). In [97] the Schwarzschild
solution was generalized to the case of n internal Ricci-flat spaces and it was shown that black hole
configuration takes place when scale factors of internal spaces are constants. In [98] an analogous
generalization of the Tangherlini solution [96] was obtained. These solutions were also generalized
to the electrovacuum case [99, 102, 100]. In [101, 100] multidimensional dilatonic black holes were
singled out. The theorem was proved in [100] that ”cuts” all non-black-hole configurations as non-
stable under even monopole perturbations. In [105] the extremely-charged dilatonic black holes
were generalized to Majumdar-Papapetrou case when the cosmological constant is non-zero.
We note that for D = 4 the pioneering Majumdar-Papapetrou solutions with conformal scalar
field and electromagnetic field were considered in [167].
At present there exists a special interest to the so-called M- [6, 8, 9]. This theory is a superme-
mbrane analogue of superstring models [31] in D = 11. The low-energy limit of these (superstring
and M -) theories leads to models governed by the Lagrangian
X θa
L = R[g] − hαβ g M N ∂M ϕα ∂N ϕβ − exp[2λa (ϕ)](F a )2 ,
a∈∆
na!
[147]. Earlier some special classes of these solutions were considered in [131, 132, 133, 150, 151, 152].
The obtained solutions take place, when certain orthogonality relations (on couplings parameters,
dimensions of ”branes”, total dimension) are imposed. In this situation a class of cosmological and
spherically-symmetric solutions was obtained [179]. Special cases were also considered in [135, 157,
158, 156]. The solutions with the horizon were considered in details in [136, 153, 154, 155, 179]. In
[155, 214] some propositions related to i) interconnection between the Hawking temperature and
the singularity behaviour, and ii) to multitemporal configurations were proved.
It should be noted that multidimensional and multitemporal generalizations of the Schwarzschild
and Tangherlini solutions were considered in [102, 172], where the generalized Newton’s formulas
in multitemporal case were obtained.
We note also that there exists a large variety of Toda-type solutions for general intersection
rules [393]. Special subclasses of these solutions (orthogonal and block-orthogonal) were obtained
in [179, 3].
In [179] (see also [164]) the Wheeler-DeWitt equation was integrated for intersecting p-branes
in orthogonal case and corresponding classical solutions were obtained also. A slightly different
approach was suggested in [159]. (For non-composite case see also [158].)
In [165, 220] exact solutions for multidimensional models with intersecting p-branes in case of
static internal spaces were obtained. Generation of the effective cosmological constant and ”infla-
tion” via p-branes was demonstrated there.
In [229, 230] the billiard behaviour was considered in multidimensional ”cosmology” with p-
branes.
We note that there exist several nice reviews devoted to certain aspects of solutions with p-
branes (see, for example, [120, 121]). However these reviews deal with more or less special classes
of p-brane solutions and their applications in superstring-, M-theories etc. Here we try to overview
more general families of solutions with composite non-localized electro-magnetic p-branes, when the
block-diagonal metrics on product manifolds are considered. The main part of the solutions under
consideration deals with Ricci-flat internal spaces of arbitrary dimensions and signatures (though
certain solutions with Einstein internal spaces are also considered).
From mathematical point of view, here we are concentrated mainly on the following key items:
(i) sigma model representation;
(ii) Toda-like Lagrangians and Toda chains;
(iii) general intersection rules for p-branes related to Cartan matrices in integrable cases.
Our approach covers more or less uniformly such topics as solutions with harmonic functions,
classical amd quantum cosmological solutions, spherically symmetric and black hole configurations,
target space symmetries, PPN asymptotics, role of frames, perfect and viscous fluids, observational
windows to extra dimensions, etc.
3. Gravitational theories with a conformal scalar field arising in different approaches [4] (they
actually can be treated as special cases of scalar-tensor theories).
4. Multidimensional unified theories in which there are dilaton fields and effective scalar fields
appearing in our 4-dimensional spacetime from additional dimensions [52,53]. They may help
also in solving the problem of a variable cosmological constant ¿from Planckian to present
values and the cosmic coincidence problem.
A striking feature of the present status of theoretical physics is that there is no satisfactory
theory unifying all four known interactions; most modern unification theories do not admit unique
and universal constant values of physical constants and of the Newtonian gravitational coupling
constant G in particular. Here we discuss various bounds that may be suggested by scalar-tensor
and multidimensional theories. Although the bounds on Ġ and G(r) are in some classes of the-
ories rather wide on purely theoretical grounds since any theoretical model contains a number of
adjustable parameters, we note that observational data concerning other phenomena, in particular,
cosmological data, may place limits on the possible ranges of these adjustable parameters.
Here we restrict ourselves to the problem of Ġ (for G(r) see [4, 12, 13, 316]). We show that
various theories predict the value of Ġ/G to be 10−12 /yr or less. The significance of this fact for
experimental and observational determinations of the value of or upper bound on Ġ is the following:
any determination with error bounds significantly below 10−12 /yr (combined with experimental
bounds on other parameters) will typically be compatible with only a small portion of existing
theoretical models and will therefore cast serious doubt on the viability of all other models. In
short, a tight bound on Ġ, in conjunction with other astrophysical observations, will be a very
effective “theory killer” and/or significantly reduce the class of viable theories. Any step forward
in this direction will be of utmost significance.
Some estimations for Ġ were done long ago in the frames of general scalar tensor theories using
the values of cosmological parameters (Ω, H, q etc) known at that time [4, 12]. It is easy to show
that for modern values they predict Ġ/G at the level of 10−12 /yr and less (see also recent estimations
of A. Miyazaki [401], predicting time variations of G at the level of 10−13 yr−1 for a Machian-type
cosmological solution in the Brans-Dicke theory).
CBPF-MO-003/02 13
The most reliable experimental bounds on Ġ/G (radar ranging of spacecraft dynamics [318]) and
laser lunar ranging [321] give the limit of 10−12 /yr), so any results at this level or less will be very
important for solving the fundamental problem of variations of constants and for discriminating
between viable unified theories. So, realization of such multipurpose new generation type space
experiments like Satellite Energy Exchange (SEE) for measuring Ġ and also absolute value of G
and Yukawa type forces at meters and Earth radius ranges [17] become extremely topical.
In what follows, we shall discuss predictions for Ġ from multidimensional models and from
generalized scalar-tensor theories, consider the influence of Ġ on the satellite orbit and the possibility
of compensation of nongravitational forces.
The model
Let us consider (4 + N )-dimensional theory described by the action
Z
1 √
Sg = 2 d4+N x −gR , (1.1)
2κ
where κ2 is the fundamental gravitational constant. Then the gravitational field equations are
T
RPM = κ2 (TPM − δPM ), (1.2)
N +2
where TPM is a (4 + N )-dimensional energy-momentum tensor, T = TM
M
, and M, P = 0, ..., N + 3.
For the (4 + N )-dimensional manifold we assume the structure
.2 . . L
2k ä 2 a N ab κ2 X α
+ + + = [ρ + (N − 1)pα3 − N pαN ], (1.7)
a2 a a2 ab N + 2 α=1
.2 . . L
b̈ b 3 ab κ2 X α
+ (N − 1) 2 + = [ρ − 3pα3 + 2pαN ]. (1.8)
b b ab N + 2 α=1
On the other hand, to get the 4-dimensional gravity equations one should put 8πG(t)ρα,(4) (t) =
κ2 ρα (t). Consequently, the effective 4-dimensional gravitational “constant” G(t) is defined by
It is worth noting that Ġ/G depends explicitly upon ḃ/b but not upon ȧ/a. Of course, it depends
upon a and ȧ/a implicitly via cosmological equations (1.6)-(1.8).
Cosmological parameters
Some inferences concerning the observational cosmological parameters of the present epoch can be
extracted directly from the equations without solving them [53]. Indeed, let us define the Hubble
parameter H, the density parameters Ωα and the ”deceleration” parameter q referring to a fixed
instant t0 in the usual way
XL
N −1 2
g −g+q− Aα Ωα = 0 (1.15)
3N α=1
with
1
Aα = [2N + 1 + 3(1 − N )ν3α + 3N νNα ], (1.16)
N +2
where
ν3α = pα3 /ρα , νNα = pαN /ρα , ρα > 0 . (1.17)
CBPF-MO-003/02 15
Note that (1.18) for N = 6, L = 1, ν31 = ν61 = 0 (so that A1 = 13/8) coincides with the
corresponding relation of Wu and Wang [50] obtained for large times in case k = −1 (see also [52]).
If k = 0, then in addition to (1.18), one can obtain a separate relation between g and Ωα , namely,
XL
N −1 2
g −g+1− Ωα = 0 (1.19)
6N α=1
(this follows from the Einstein equation R00 − 21 R = κ2 T00 , which is certainly a linear combination
of (1.6)-(1.8).
The present observational upper bound on g is
<
| g |∼ 0.1 (1.20)
and the recent value of the Hubble parameter H = (0.7 ± 0.1) × 10−10 (y −1 ) ≈ 70 ± 10 (km/s.Mpc).
Also, it is necessary to cite here Damour-Taylor binary pulsar result [319] that gives slightly
different bounds on G-dot.
In what follows, to illustrate the multidimensional model predictions for Ġ, we suggest an
example of a 2-component model in dimension D = 10 with negative q (i.e. positive acceleration)
and g obeying (1.20).
It was mentioned in [229] that the multidimensional cosmological model on product manifold
R × M1 × ... × Mn with fields of forms (for review see [3, 437]) may be decribed in terms of
multicomponent ”perfect” fluid [72] with the following equations of state for α-s component: pαi =
−ρα if p-brane worldvolume contains Mi and pαi = ρα in opposite case. Thus, the field of form
matter leads us either to Λ-term, or to stiff matter equations of state in internal spaces. [We remind
here that the action with the fields of forms (and without dilatonic fields) reads
Z p X 1
S = dD z |g|{R[g] − (F a )2 }, (1.24)
a
n a !
CBPF-MO-003/02 16
where g = gM N dz M ⊗ dz N is the metric, F a = dAa is na -form. In the p-brane ansatz the forms F a
are chosen as sums of monoms proportional to forms of the type τi1 ∧ .... ∧ τik , where τi is volume
form on Mi corresponding to the metric g i .]
In this case, calculating Aα for a two-component fluid from (1.16), we get from (1.18) for small
g
2N + 1 1 N −1 2
g≈q− Ω +4 Ω, (1.25)
N +2 N +2
and for a spatially flat isotropic model with k = 0 and small g we obtain from (1.19)
1 − g ≈ Ω1 + Ω2 . (1.26)
Now we illustrate the formulas by the following example when the number of extra dimensions
N is 6 (K 6 may be a Calabi-Yau manifold) and
−q = Ω1 = Ω2 = 0.5. (1.27)
We get from (1.25)
1
g≈− ≈ −0.06 (1.28)
16
in agreement with the present experimental bounds on G (??). This implies for Ġ:
In this case the second fluid component corresponds to a magnetic (Euclidean) 5-brane.
Here R[g] is the scalar curvature, g = | det(gµν )|; f, h and U are certain functions of φ, varying
from theory to theory, Lm is the matter Lagrangian.
This formulation of the theory corresponds to the Jordan conformal frame, in which matter par-
ticles move along geodesics and hence the weak equivalence principle is valid, and non-gravitational
fundamental constants do not change. In other words, this is the frame well describing the existing
laboratory, geophysical and cosmological observations.
Among the three functions of φ entering into (1.30) only two are independent since there is a
freedom of transformations φ = φ(φnew ). We use this arbitrariness, choosing h(φ) ≡ 1, as is done,
e.g., in [322]. Another standard parametrization is to put f (φ) = φ and h(φ) = ω(φ)/φ (the Brans-
Dicke parametrization of the general theory (1.30)). In our parametrization h ≡ 1, the Brans-Dicke
parameter ω(φ) is ω(φ) = f (fφ )−2 ; here and henceforth, the subscript φ denotes a derivative with
respect to φ. The Brans-Dicke STT is the particular case ω = const, so that in (1.30)
φ − 12 R fφ + Uφ = 0, (1.32)
³ ´
f (φ) Rνµ − 12 δµν R = −φ,µ φ,ν + 12 δµν φ,α φ,α − δµν U (φ) + (∇µ ∇ν − δµν )f − Tµν (m) , (1.33)
where is the D’Alembert operator, and the last term in (1.33) is the energy-momentum tensor of
matter.
Consider now isotropic cosmological models with the standard FRW metric
h dr2 i
2 2 2 2 2 2 2
ds = dt − a (t) + r (dθ + sin θ dϕ ) , (1.34)
1 − kr2
where a(t) is the scale factor of the Universe, and k = 1, 0, −1 for closed, spatially flat and
hyperbolic models, respectively. Accordingly, we assume φ = φ(t) and the energy-momentum
tensor of matter in the perfect fluid form Tµν (m) = diag(ρ, −p, −p, −p) (ρ is the density and p is the
presuure).
The field equations in this case can be written as follows:
ȧ 3
φ̈ + 3 φ̇ − 2 (aä + ȧ2 + k) + Uφ = 0, (1.35)
a a
3f 2 ȧ
2
(ȧ + k) = 12 φ̇2 + U − 3 f˙ + ρ, (1.36)
a a
f 2 1 2 ¨ − 2 ȧ f˙ − p.
(2aä + ȧ + k) = − 2
φ̇ + U − f (1.37)
a2 a
To connect these equations with observations, let us fix the time t at the present epoch (i.e.,
consider the instantaneous values of all quantities) and introduce the standard observables:
H = ȧ/a (the Hubble parameter),
q = −aä/ȧ2 (the deceleration parameter),
Ωm = ρ/ρcr (the matter density parameter),
where ρcr is the critical density, or, in our model, the r.h.s. of Eq. (1.36) in case k = 0: ρcr = 3f H 2 .
This is slightly different from the usual definition ρcr = 3H 2 /8πG where G is the Newtonian
gravitational constant. The point is that the locally measured Newtonian constant in STT differs
from 1/(8πf ); provided the derivatives Uφφ and fφφ are sufficiently small, one has [322]
1 2ω + 4
8πGeff = . (1.38)
f 2ω + 3
Since, according to the solar-system experiments, ω ≥ 2500, for our order-of-magnitude reasoning
we can safely put 8πG = 1/f , and, in particular, our definition of ρcr now coincides with the
standard one.
The time variation of G, to a good approxiamtion, is
where, for convenience, we have introduced the coefficient g expressing Ġ/G in terms of the Hubble
parameter H.
Eqs. (1.35)-(1.37) contain too many arbitrary parameters for making a good estimate of g. Let
us now introduce some restrictions according to the current state of observational cosmology:
(i) k = 0 (a spatially flat cosmological model, so that the total density of matter equals ρcr );
CBPF-MO-003/02 18
(ii) p = 0 (the pressure of ordinary matter is negligible compared to the energy density);
(iii) ρ = 0.3 ρcr (the ordinary matter, including its dark component, contributes to only 0.3 of
the critical density; unusual matter, which is here represented by the scalar field, comprises the
remaining 70 per cent).
Then Eqs. (1.36) and (1.37) can be rewritten in the form
1 2
2
φ̇ + U − 3H f˙ = 2.1H 2 f, (1.40)
− 21 φ̇2 + U − 2H f˙ − f¨ = (1 − 2q)H 2 f. (1.41)
Subtracting (1.41) from (1.40), we exclude the “cosmological constant” U , which can be quite large
but whose precise value is hard to estimate. We obtain
φ̇2 − H f˙ + f¨ = (1.1 + 2q)H 2 f. (1.42)
The first term in Eq. (1.42) can be represented in the form
φ̇2 = f˙2 (df /dφ)−2 = f˙2 ω/f,
and f˙/f can be replaced with −gH. The term f¨ can be neglected for our estimation purposes. To
see this, let us use as an example the Brans-Dicke theory, in which f = φ2 /(4ω). We then have
f¨ = (φ̇2 + φφ̈)/(2ω);
here the first term is the same as the first term in Eq. (1.42), times the small parameter 1/(2ω).
Assuming that φφ̈ is of the same order of magnitude as φ̇2 (or slightly greater), we see that,
generically, |f¨| ¿ φ̇2 . Note that our consideration is not restricted to the Brans-Dicke theory and
concerns the model (1.30) with an arbitrary function f (φ) and an arbitrary potential U (φ).
Neglecting f¨, we see that (1.42), divided by H 2 f , leads to a quadratic equation with respect to
g:
ωg 2 + g − q 0 = 0, (1.43)
where q 0 = 1.1 + 2q.
According to modern observations, the Universe is expanding with an acceleration, so that the
parameter q is, roughly, −0.5 ± 0.2, hence we can take |q 0 | ≤ 0.4.
In case q 0 = 0 we simply obtain g = −1/ω. Assuming
H = h100 · 100 km/(s.Mpc) ≈ h100 · 10−10 yr−1
and ω ≥ 2500, we come to the estimate
|Ġ/G| ≤ 4·10−14 h100 yr−1 , (1.44)
where h100 is, by modern views, close to 0.7. So (1.44) becomes
|Ġ/G| ≤ 3·10−14 yr−1 . (1.45)
For nonzero valuespof q 0 , solving the quadratic equation (1.43) and assuming q 0 ω À 1, we arrive
at the estimate |g| ∼ q 0 /ω, so that, taking q 0 = 0.4 and again ω ≥ 2500, we have instead of (1.44)
|Ġ/G| ≤ 1.3·10−12 h100 yr−1 ≈ 0.9·10−12 yr−1 , (1.46)
where we have again put h100 = 0.7.
We conclude that, in the framework of the general STT, modern cosmological observations,
taking into account the solar-system data, restrict the possible variation of G to values within
10−12 /yr. This estimate may be considerably tightened if the matter density parameter Ωm and the
(negative) deceleration parameter q will be determined more precisely.
CBPF-MO-003/02 19
1.3.4 Estimation of satellite orbit dependence on the time variations of G on the level
Ġ/G = 10−13 per year
The influence of small G variation may be considered as a small perturbation of the satellite orbit.
Such variation may be expressed as
G = G0 + Ġt (1.47)
where G0 is the present value of the gravitational constant and Ġ its time derivative. Consider for
simplicity the case of a plane circular orbit. In this case the polar coordinates of the satellite will
be
r = r0 + δr, ϕ = ωt + δϕ (1.48)
p
where r0 and ω = G0 ME /r03 are the unperturbed radius of the orbit and angular velocity, ME is
the Earth mass. The Lagrangian of planar motion has the form
m¡ 2 ¢ GME m
L= ṙ + r2 · ϕ̇2 + (1.49)
2 r
After substitution of (1.47) and (1.48) into (1.49) and standard variation with respect to δr and δϕ
the following equations for the perturbed circular motion will be obtained:
¡ ¢ G0 ME (1 + Ġ t) G0 ME
δr̈ − δr ω 2 + 2ωδ ϕ̇ + δ ϕ̇2 − r0 (2ωδ ϕ̇ + δ ϕ̇2 ) + G
− =0 (1.50)
(r0 + δr)2 r02
2δ ṙδ ϕ̇ 2ωδ ṙ
δ ϕ̈ + + =0 (1.51)
r0 + δr r0 + δr
A numerical solution of the system (1.50)–(1.51) for the orbit altitude Horb = 1500 km with Ġ/G =
10−13 year−1 (or 3.17098 · 10−21 s−1 ) and zero initial conditions for δr and δϕ gives the following
values of δr and δϕ during a year: δr ' −7.9·10−7 m and δϕ ' 2.84·10−9 . The linear displacement
of the satellite along its orbit with respect to its unperturbed motion (corresponding to Ġ = 0) is
equal to δl ' r0 · δϕ = 0.022 m/year.
The above estimate coincides with the estimates obtainable using the adiabatic invariant method.
Namely, if the gravitational constant G slowly varies, then it follows from the adiabatic invariant
method that the orbital radius r (the large semiaxis in the general case) and the orbital period T
vary as
δr Ġ δT Ġ
=− , = −2 , (1.52)
r G T G
and for Ġ/G = 1 · 10−13 year−1 and r ≈ 7.9 · 106 m (which corresponds to the orbital altitude
Horb ≈ 1500 m) Eq. (1.52) gives the estimate δr ≈ 10−7 ÷ 10−6 m/year.
Thus we see, that there are two possibilities for measuring Ġ/G with an accuracy Ġ/G =
1·10−13 /yr. First, one can measure the changes in the orbit radius or satellite position in the orbit.
For this purpose it is necessary to predict the orbital radius (or its large semiaxis) up to ∼ 7 · 10−6
m/year and the satellite position in the orbit up to 0.022 m/year. To do that it is necessary to
know the initial value of the orbital radius up to 10−7 m, while the existing methods give only a
10−2 m accuracy. It means that measuring the orbital radius or the satellite position does not give
a constructive tool for measuring Ġ/G up to 1 · 10−13 /yr.
Second, instead of measuring the position, we can try to measure changes of the orbital period
T of the satellite. To determine the period up to dT /T ∼ 10−13 /yr during 1 revolution we must fix
CBPF-MO-003/02 20
the position up to 5·10−4 cm (or an equivalent angle). If we can fix the satellite position on the orbit
up to 1 cm, then after approximately 0.5 year of measurements we can determine the period with
the required accuracy. After that the measurement of the orbital period T must be repeated some
time later (≥ 1 year). During each of these two series of measurements it is necessary to fix position
of some particular point of the orbit up to 1 cm (or an equivalent angular coordinate accuracy)
during the measurement period, i.e., 0.5 year. It is an extremely nontrivial problem, because there
is a great difference between the possibility to determine the position (coordinates) of a satellite
up to 1 cm and to fix some marked point of the orbit for half a year with the same accuracy. Lastly,
it is also necessary to recall that the real orbital motion of a satellite is not periodic, since a real
orbit is not closed and a correct definition of orbital period with a required accuracy is a nontrivial
problem.
The capacity sensors of the drag-free system have a ponderomotive action (“back action”) on the
shepherd during a short time of the order of 1 s. According to [324], “the mass of the SEE satellite
is to be distributed so that it exerts no gravitational force on the test bodies within”. Let the test
body, the shepherd, move freely (without any influence from the satellite) and pass a distance of 50
cm between two contacts with the capacity sensors. The free motion time can be estimated using
Eq. (1.53) for the acceleration aL — see (1.58):
In this time interval, the average acceleration of the satellite due to the back action of the drag-free
system is equal to
τ
aE = 7.5 · 10−15 g = 1.97 · 10−18 g (1.60)
tf
If there is an adequate model of the Earth’s gravitational field, then the drag-free system anal-
ogous to that in TRIAD can compensate non-gravitational forces so that it can predict the SEE
trajectory (shepherd trajectory) with an accuracy of 1 cm per year. It is possible due to a large
volume protected from the gravitational field of the satellite.
• The gravitational perturbations due to mass defects in the capsule apparently create the main
difficulties. Because of a large size of the proof body, the shepherd ((∼ 40 cm), the influence
of “high-frequency” (at small space distances) mass defects will be averaged and decreased.
• Braginsky and Manukin [325] estimated an effect of the magnetic field of the Earth on satellite
motion and obtained that the effect can be neglected.
• The angular resolution of 3.15·10−14 per year (according to the formula L5 [324]) allows one to
test the existence of new long-range forces (inverse square law) with a resolution with respect
to the parameter α of 9,598·10−17 for λ = 3.935·106 m.
• With the same angular resolution the Lense-Thiring effect can be tested with an accuracy of
10−6 -10−7 . Recently Ciufolini tested the effect with an accuracy of 20% [326].
1.3.6 Conclusion
Summarizing the above considerations, we can conclude:
• Restrictions of possible nonzero values of Ġ give no restriction on the possible class of gen-
eralized gravitation theories, but in the framework of some fixed theory any restriction on Ġ
restricts the possible class of models.
• The possibility of measuring of Ġ/G = 1·10−13 year−1 using the induced changes of the orbital
parameters or satellite displacement along its orbit seems rather difficult. One of the main
problems may be the knowledge of the gravitational field of the Earth with reqired precision.
CBPF-MO-003/02 22
• The use of a SEE drag-free satellite make it possible in principle to perform several high
precision tests, such as verification of the existence of new long-range forces (inverse square
law) with a resolution of the parameter α up to 9.598·10−17 for λ = 3.935·106 m, verification of
the Lense-Thiring effect with an accuracy of 10−6 -10−7 and for construction of a more accurate
model of the Earth’s gravitational field.
CBPF-MO-003/02 23
2 The model
2.1 The action and equations of motion
We consider the model governed by an action
Z p
1 D
S= 2
d z |g|{R[g] − 2Λ − hαβ g M N ∂M ϕα ∂N ϕβ (2.1)
2κ M
X θa
− exp[2λa (ϕ)](F a )2g } + SGH , (2.2)
a∈∆
na!
1 a
F a = dAa = F dz M1 ∧ . . . ∧ dz Mna (2.3)
na ! M1 ...Mna
is a na -form (na ≥ 2) on a D-dimensional manifold M , Λ is a cosmological constant and λa is a
1-form on Rl : λa (ϕ) = λaα ϕα , a ∈ ∆, α = 1, . . . , l. In (7.1305) we denote |g| = | det(gM N )|,
(F a )2g = FM
a
1 ...Mna
FNa 1 ...Nna g M1 N1 . . . g Mna Nna , (2.4)
a ∈ ∆, where ∆ is some finite set, and SGH is the standard Gibbons-Hawking boundary term [169].
In the models with one time all θa = 1/2 when the signature of the metric is (−1, +1, . . . , +1).
The equations of motion corresponding to (7.1305) have the following form
1
RM N − gM N R = TM N − ΛgM N , (2.5)
2
X λα
4[g]ϕα − θa a e2λa (ϕ) (F a )2g = 0, (2.6)
a∈∆
na !
∇M1 [g](e2λa (ϕ) F a,M1 ...Mna ) = 0, (2.7)
a ∈ ∆; α = 1, . . . , l. In (7.1308) λαa = hαβ λβa , where (hαβ ) is matrix inverse to (hαβ ). In (3.273)
X
TM N = TM N [ϕ, g] + θa e2λa (ϕ) TM N [F a , g], (2.8)
a∈∆
where
µ ¶
α 1β α P β
TM N [ϕ, g] = hαβ ∂M ϕ ∂N ϕ − gM N ∂P ϕ ∂ ϕ , (2.9)
2
1 1 a,M2 ...Mna
TM N [F a , g] = a
[− gM N (F a )2g + na FM M2 ...Mna FN ]. (2.10)
na ! 2
In (7.1308) and (7.1410) operators 4[g] and 5[g] are Laplace-Beltrami and covariant derivative
operators, respectively, corresponding to g.
CBPF-MO-003/02 24
where g 0 = gµν 0
(x)dxµ ⊗ dxν is an arbitrary metric with any signature on the manifold M0 and
g = gmi ni (yi )dyi ⊗ dyini is a metric on Mi satisfying the equation
i i mi
Rmi ni [g i ] = ξi gm
i
i ni
, (2.13)
where di is both the dimension of the oriented manifold Mi and the rank of the volume form τi . We
also put τ (∅) = ε(∅) = 1 and d(∅) = 0.
In the Appendix 1 we outline for completeness all relations for Riemann and Ricci tensors
corresponding to the metric (7.1415).
For fields of forms we consider the following composite electromagnetic ansatz
X X
Fa = F (a,e,I) + F (a,m,J) (2.20)
I∈Ωa,e J∈Ωa,m
where
are elementary forms of electric and magnetic types respectively, a ∈ ∆, I ∈ Ωa,e , J ∈ Ωa,m and
Ωa,v ⊂ Ω, v = e, m. In (2.22) ∗ = ∗[g] is the Hodge operator on (M, g)
|g|1/2
(∗ω)M1 ...MD−k = εM1 ...MD−k N1 ...Nk ω N1 ...Nk , (2.23)
k!
CBPF-MO-003/02 25
for some i, j ∈ w1 , i 6= j.
Let us define I¯ as follows
I¯ ≡ I0 \ I, I0 = {1, . . . , n}. (2.29)
Restriction 2 (only for d0 = 1, 3). For any a ∈ ∆ there are no I ∈ Ωa,e and J ∈ Ωa,m such that
J¯ = {i} t I f or d0 = 1, (2.30)
I = {i} t J¯ f or d0 = 3, (2.31)
where i ∈ w1 .
Restriction 1 is satisfied for n1 ≤ 1 and also in the non-composite case: |Ωa,e | + |Ωa,m | = 1
for all a ∈ ∆. For n1 ≥ 2 it forbids the following pairs of two electric or two magnetic p-branes,
corresponding to the same form F a , a ∈ ∆:
i I
j J
CBPF-MO-003/02 26
Figure 1.
i J¯ i I
I J¯
Figure 2.
dF s = 0, (2.34)
s ∈ Sm , for fields from (7.1415), (2.20)–(2.24), when Restrictions 1 and 2 are imposed, are equivalent
to equations of motion for the σ-model governed by the action (see also Proposition 0 below )
Z p n
1
Sσ0 = d x |g 0 | R[g 0 ] − ĜAB g 0µν ∂µ σ A ∂ν σ B
d0
(2.35)
2κ20
X o
− εs exp (−2UAs σ A )g 0µν ∂µ Φs ∂ν Φs − 2V , (2.36)
s∈S
is the potential, µ ¶
Gij 0
(ĜAB ) = , (2.38)
0 hαβ
is the target space metric with
di dj
Gij = di δij + , (2.39)
d0 − 2
and co-vectors
X
UAs = UAs σ A = di φi − χs λas (ϕ), (UAs ) = (di δiIs , −χs λas α ), (2.40)
i∈Is
CBPF-MO-003/02 27
Equations of motion corresponding to the action (2.36) with the potential (2.37) have the fol-
lowing form
X 2V 0
Rµν [g 0 ] = ĜAB ∂µ σ A ∂ν σ B + εs exp (−2UAs σ A )∂µ Φs ∂ν Φs + gµν , (2.45)
s∈S
d0−2
X ∂V
ĜAB ∆[g 0 ]σ B + εs UAs exp (−2UCs σ C )g 0µν ∂µ Φs ∂ν Φs = A
, (2.46)
s∈S
∂σ
³p ´
0 0µν s A s
∂µ |g |g exp (−2UA σ )∂ν Φ = 0, (2.47)
i, j ∈ w1 , where
X X
Cij = θa exp[2λa (ϕ)]Cij (Φa,e , φ, g 0 ) − ε[g] θa exp[−2λa (ϕ)]Cij (Φa,m , φ, g 0 ), (2.49)
a∈∆ a∈∆
X
Cij (Φa,v , φ, g 0 ) = δ(i, I ∩ J)δ(j, I ∩ J)ε(I ∩ J)
(I,J)∈Wij (Ωa,v )
(2.50)
³ X ´
× exp −2 dl φl g 0µν ∂µ Φ(a,v,I) ∂ν Φ(a,v,J) , (2.51)
l∈I∩J
They are non-trivial if there are pairs of electric or magnetic p-branes with the same ”colour” index
a and intersection depicted at Figure 1.
The second group of constraints (”electro-magnetic” ones) appear only if d0 = 1, 3 (here d0 6= 2)
and n1 = |w1 | ≥ 1. They read
(d )
Ciµ 0 = 0, (2.55)
i ∈ w1 and µ = 1, . . . , d0 .
Here for d0 = 1 we denote
(1)
X X
Ci1 = µ(J)δ(j, I)∂1 Φ(a,e,I) ∂1 Φ(a,m,J) , (2.56)
a∈∆ (1)
(I,J)∈Wia
where
Wia ≡ {(I, J) ∈ Ωa,e × Ωa,m |J¯ = {j} t I},
(1)
(2.57)
i ∈ w1 , and µ(J) = ±1 is defined by relation
¯ ∧ dxµ ∧ τ (J).
µ(J)dxµ ∧ τ (I0 ) = τ (J) (2.58)
(3) ¯
Wia ≡ {(I, J) ∈ Ωa,m × Ωa,e |J = {i} t I}, (2.60)
i ∈ w1 and µ = 1, 2, 3, where ∇ρ Φ = g 0ρν ∂ν Φ.
These formulas follows from the relation for the off-block-diagonal components of energy-momentum
tensor:
³ X n ´
j 1/2 0 −1/2 (1)
T1j 10 = ε[g]|g | |g | exp −3γ − di φi + 2φj Cj1 , (2.61)
i=1
for d0 = 1, and
³ n
X ´
i 1/2 (3)
Tµ0 1i = ε[g0 ]ε[g]|g | exp −γ − di φi Ciµ ; (2.62)
i=1
d0 6= 2 and γ = γ0 (φ) from (2.27) the equations of motion (3.273)-(7.1410) and Bianchi identities
(2.34) are equivalent to the equations of motion for the σ-model (2.36) and the constraints (2.48)
(for all d0 6= 2 ) and (2.55) (for d0 = 1, 3) imposed.
Proof. The appearance of constraints was verified above. Now we consider the reduction
to σ-model itself. For (F, ϕ)-part of field equations and Bianchi identities the equivalence with
corresponding equations of motion for σ-model (see eqs. (2.46) and (2.47)) can be readily verified.
Here we consider the Einstein equations (3.273) written in the form
2Λ
R M N = ZM N + gM N , (2.63)
D−2
where
T
ZM N ≡ T M N + gM N , (2.64)
2−D
and T = TM M . Here X
ZM N = ZM N [ϕ] + θa e2λa (ϕ) ZM N [F a , g], (2.65)
a∈∆
where
For block-diagonal part of (2.67) we have (see relations from Appendix 2),
X X
ZM N [F a , g] = ZM N [F (a,e,I) , g] + ZM N [F (a,m,J) , g], (2.68)
I∈Ωa,e J∈Ωa,m
where (M, N ) = (µ, ν), (mi , ni ); i = 1, . . . , n (F (a,e,I) and F (a,m,J) are defined in (??) and (2.22),
respectively).
Using the relations for Ricci tensor ¿from Appendix 1 with γ = γ0 from (2.27) and relations
from Appendix 2 we obtain that (mi , ni )-components of Einstein equations (2.63)(i = 1, . . . , n)
are equivalent to φi -part of σ-model equations (2.46) and (µ, ν)-components of Einstein equations
(2.63) are equivalent to σ-model Einstein equations (??). Note that dealing with (µ, ν)-components
of (2.63) we use the relation for γ = γ0 (φ) with φ substituted from (mi , ni )-equations. Also the
following relations should be used:
(see formulas (12.1954) and (12.1955) from Appendix 2) and the relations for the contravariant
components of U s -vectors defined in (2.40), namely U sA = ĜAB UBs with the matrix (ĜAB ) inverse
to one from (2.38) [147]:
d(Is )
U si = δiIs − , U sα = −χs λαas , (2.70)
D−2
s = (as , vs , Is ). Proposition is proved.
CBPF-MO-003/02 30
where n
X
f = f (γ, φ) = (d0 − 2)γ + dj φj . (2.73)
j=1
For finite internal space volumes Vi (e.g. compact Mi ) and electric p-branes (i.e. all Ωa,m = ∅)
the action (2.72) coincides with the action (7.1305) when
n
Y
κ2 = κ20 Vi . (2.74)
i=1
This may be readily verified using the relation (12.1917) from Appendix 2 and the scalar curvature
decomposition (see also Appendix 1)
n
X n n
X
−2φi i −2γ 0
R[g] = e R[g ] + e R[g ] − di (∂φi )2 (2.75)
i=1 i=1
o
− (d0 − 2)(∂γ)2 + (∂f )∂(f + 2γ) + RB , (2.76)
where
p p
RB = (1/ |g 0 |)e−f ∂µ [−2ef |g 0 |g 0 µν
∂ν (f + γ)] (2.77)
I¯ = J (2.79)
or J¯ = {i} t (J¯ ∩ I), I = (J¯ ∩ I) t {j} (2.80)
for some i, j ∈ w1 , i 6= j.
CBPF-MO-003/02 31
s ∈ S, satisfies to field equations (??)–(2.47) with V = 0 if (real) numbers νs obey the relations
X 0
(U s , U s )εs0 νs20 = −1 (3.89)
s0 ∈S
CBPF-MO-003/02 32
where i = 1, . . . , n, α = 1, . . . , l, a ∈ ∆ and
= se,= νs (∗0 dHs ) ∧ τ (I¯s ), for vs = m,
F s = νs dHs−1 ∧ τ (Is ), for vs F (3.97)
Hs are harmonic functions on (M0 , g 0 ) coinciding inside blocks (see (??)) and relations (3.90) on
0
parameters νs are imposed. Here the matrix ((U s , U s )) and parameters εs , s ∈ S, are defined
in (3.85) and (7.1313), respectively; λαa = hαβ λβa , ∗0 = ∗[g 0 ] is the Hodge operator on (M0 , g 0 )
and I¯ is defined (2.29). In deriving the solution from Proposition 1 the relations for contravariant
components of U s -vectors (2.70) were used. Relation (3.97) was obtained from (2.22) by use of the
following relations from [147]
X
F (a,m,J) = ε(J)µ(J) exp(−2 ¯
di φi − 2λa (ϕ))(∗0 dΦ(a,m,J) ) ∧ τ (J), (3.98)
i∈J
à !
X
i 2
exp −2λa (ϕ) − 2 di φ = H(a,m,J) . (3.99)
i∈J
where µ(J) = ±1 is defined in (2.58). The relation (3.99) is a special case of a more general identity
exp(−2UAs σ A ) = Hs2 , s ∈ S, (3.100)
CBPF-MO-003/02 33
following from Proposition 1. In (3.97) we redefined the sign of νs -parameter (compared to (2.22))
as following: νs 7→ −ε(I)µ(I)νs .
Remark 1. The solution (??)-(3.97) is also valid for d0 = 2, if Restriction 2 from previous
section is replaced by Restriction 2∗ . It may be verified using the sigma-model representation
(2.72).
For pseudo-Euclidean metric g all ε(Is ) = −1 and, hence, all p-branes should contain time manifold.
For the metric g with the Euclidean signature only magnetic p-branes can exist in this case.
From scalar products (3.85) and the orthogonality condition (3.101) we get the ”orthogonal”
intersection rules
d(Is )d(Is0 )
d(Is ∩ Is0 ) = − χs χs0 λas α λas0 β hαβ ≡ ∆(s, s0 ), (3.108)
D−2
for s = (as , vs , Is ) 6= s0 = (as0 , vs0 , Is0 ).
Example 1: D = 11 supergravity [29]. The action for the bosonic sector of D = 11
supergravity with omitted Chern-Simons term has the following form
Z p 1
Ŝ11 = d11 z |g|{R[g] − F 2 }. (3.109)
M 4!
Here rankF = 4 and the signature of g is (−, +, . . . , +).
The dimensions of p-brane worldvolumes are (see (??))
6, for vs = m. (3.111)
CBPF-MO-003/02 34
The model describes electrically charged 2-branes and magnetically charged 5-branes.
From (3.108) we obtain the intersection rules [132]
d(Is ∩ Is0 ) = 1, for vs = vs0 = e;
2, for vs = e, vs0 = m; (3.112)
4, for vs = vs0 = m.,
The Restrictions 1 and 2 from Section 2 are satisfied in this case (see also (2.32) and (2.33). Here
(U s , U s ) = 2, νs2 = 1/2 and
εs = ε(Is ) = −1 (3.113)
for all s ∈ S, i.e. all p-branes are pseudo-Euclidean.
The solution (??)-(3.97) reads
n n
X o
0
g = Ue Um ĝ + Ui ĝ i , (3.114)
i=1
³Y ´ 13 ³ Y ´ 23
Ue = Hs , Um = Hs , (3.115)
s∈Se s∈Sm
Y
Ui = Hs−δiIs , (3.116)
s∈S
X X
F = νs dHs−1 ∧ τ (Is ) + νs (∗0 dHs ) ∧ τ (I¯s ), (3.117)
s∈Se s∈Sm
where ∗0 dH is the Hodge dual form on (M0 , g 0 ). The metric and fields of forms are defined on the
manifold (7.1310) and all functions Hs , s ∈ S are harmonic on (M0 , g 0 ). The subsets Ωa,v ∈ Ω,
v = e, m, satisfy the intersection rules (3.112) and the signature restrictions (3.113).
The solutions also satisfy the equations of motion with Chern-Simon term taken into account.
This can be readily verified using the relation for the bosonic part of action for D = 11 supergravity
[29] Z
S11 = Ŝ11 + c11 A∧F ∧F (3.118)
M
where c11 = const and Ŝ11 is defined in (3.109) (F = dA). Indeed, the only modification of equations
of motion is related to ”Maxwell” equation
d ∗ F = const F ∧ F, (3.119)
with F ∧ F = 0 in the solutions under consideration. These solutions [143] coincide with those
obtained in [149, 132]) for flat (Mν , g ν ), ν = 0, . . . , n.
s, s0 ∈ S. Here some ordering in S is assumed. Using this definition and (3.85) we obtain the
intersection rules
1
d(Is ∩ Is0 ) = ∆(s, s0 ) + Ks0 Ass0 (3.122)
2
s 6= s0 , where ∆(s, s0 ) is defined in (3.108).
For det A 6= 0 relation (3.89) may be rewritten in the equivalent form
X 0
−εs νs2 (U s , U s ) = 2 Ass ≡ bs , (3.123)
s0 ∈S
0
s ∈ S, where (Ass ) = A−1 . Thus, eq. (3.89) may be resolved in terms of νs for certain εs = ±1,
s ∈ S. We note that due to (3.87) the matrix A has a block-diagonal structure and, hence, for
any i-th block the set of parameters (νs , s ∈ Si ) depend upon the matrix inverse to the matrix
(Ass0 ; s, s0 ∈ Si ).
Now we consider one-block case when the p-brane intersections are related to some Lie algebras.
3.1.2.1. Finite dimensional Lie algebras [217].
Let A be a Cartan matrix of a simple finite-dimensional Lie algebra. In this case
Ass0 ∈ {0, −1, −2, −3}, s 6= s0 . The elements of inverse matrix A−1 are positive (see Ch. 7 in [187])
and hence we get from (3.123) the same signature relation (3.103) as in orthogonal case. Moreover,
all bs are natural numbers:
bs = ns ∈ N, (3.124)
s ∈ S. Integers ns coincide with the components of twice the so-called dual Weyl vector in the basis
of simple coroots (see Ch.3.1.7 in [187]). Explicit formulas for ns , corresponding to simple finite
dimensional Lie algebras are outlined in Appendix 3.
Here we consider three examples of solutions in D = 11 and D = 10 (IIA) supergravities and
so-called BD -models (see Example 4 below) corresponding to A2 -algebra with the Cartan matrix
µ ¶
2 −1
A= (3.125)
−1 2
and n1 = n2 = 2.
Example 2: A2 -dyon in D = 11 supergravity. For the D = 11 supergravity with the
bosonic part of the action (3.109) we get for A2 -solutions with two branes the intersections
2 ∩ 5 = 1, 5 ∩ 5 = 3. (3.126)
The electromagnetic dyon (i.e. the bound state of electric and magnetic branes) with the intersection
2 ∩ 5 = 1 reads
g = H 2 ĝ 0 − H −2 dt ⊗ dt + ĝ 1 + ĝ 2 , (3.127)
F a = ν1 dH −1 ∧ dt ∧ τ1 + ν2 (∗0 dH) ∧ τ1 , (3.128)
the 4-dimensional section of the metric (??) coincides with the metric of Majumdar-Papapetrou
solution [166] describing N extremal charged black holes with horizons at points xi , and charges
qi > 0, i = 1, . . . , N . The solution (??)–(3.128) with H from (3.129) describes N extremal p-brane
dyonic black holes. Any dyon contains one electric ”brane” and one magnetic ”brane” with equal
charge densities.
Example 3: A2 -dyon in IIA supergravity. The bosonic part of action for D = 10 IIA
supergravity reads
Z p n X4 o 1Z
10 2 2λa ϕ a 2
S= d z |g| R[g] − (∂ϕ) − e (F ) − F 4 ∧ F 4 ∧ A2 , (3.130)
a=2
2
for a = 2, 3, 4, respectively.
We consider here the sector corresponding to a = 3, 4 describing electric p-branes: fundamental
string (FS), D2-brane and magnetic p-branes: N S5- and D4-branes. We get (U s , U s ) = 2 for all s.
The solutions with A2 intersection rules corresponding to relations
are valid in the ”truncated case” (without Chern-Simons term) and in a general case (3.130) as
well. Here (p1 ∩ p2 = d) ⇔ (d(I) = p1 + 1, d(J) = p2 + 1, d(I ∩ J) = d).
Let us consider the solution describing the electromagnetic dyon with intersections
1 ∩ 5 = 2 ∩ 4 = 1. (3.134)
The solution is given by relations (??) and (3.128) with H being harmonic function on (M0 , g 0 ),
d0 = 3, ν12 = ν22 = 1, d1 = a − 2, d2 = 8 − a, a = 3, 4, and metrics g ν having Euclidean signatures.
Example 4: A2 -dyon in BD -models. Now we consider examples of solutions for BD -models
with the action [179]
Z p n D−7
X 1 o
SD = dD z |g| R[g] + g M N ∂M ϕ
~ ∂N ϕ
~− exp[2~λa ϕ
~ ](F a )2 , (3.135)
a=4
a!
a, b = 4, . . . , D − 7 and ~λD−7 = −2~λ4 . For D > 11 vectors ~λ4 , . . . , ~λD−8 are linearly independent.
The model (3.135) contains l scalar fields with a negative kinetic term (i.e. hαβ = −δαβ in
(7.1305)) coupled to (l + 1) forms. For D = 11 (l = 0) the model (3.135) coincides with the
CBPF-MO-003/02 37
Thus, there are (l + 1) electric and (l + 1) magnetic p-branes, p = d(I) − 1. In BD -model all Ks = 2.
We consider BD -model, D ≥ 11. Let a ∈ {4, . . . , D−7}, g 3 = −dt⊗dt, d1 = a−2, d2 = D−2−a,
d0 = 3 and metrics g 0 , g 1 , g 2 are Ricci-flat. The A2 -solution describing a dyon configuration with
electric d1 -brane and magnetic d2 -brane, corresponding to F a -form and intersecting in 1-dimensional
time manifold reads as given by relations (??), (3.128) and ϕ ~ = 0, where H is the harmonic function
on (M0 , g 0 ) and ν12 = ν22 = 1. The case D = 11 was considered in Example 2. For D = 12 we have
two possibilities: a) a = 4, d1 = 2, d2 = 6; b) a = 5, d1 = 3, d2 = 5. The signature restrictions on
g 1 and g 2 are the following: ε1 = +1, ε2 = −ε[g]. They are satisfied when g 0 and g 1 are metrics of
Euclidean signature.
Remark 3. In Examples 2, 3 and 4 the A2 -dyon solutions do not satisfy the Restriction 2
(or, equivalently, (2.33)) that guarantees the vanishing of non-block-diagonal components of stress-
energy tensor, i.e. T1i µ0 = 0, (µ = 1, 2, 3, di = 1). Nevertheless this vanishing does take place since
(3)
Ciµ = 0 due to Φ(a,m,J) = Φ(a,e,I) (see (2.59) and (2.62).
εs (U s , U s ) > 0, (3.139)
0
s ∈ Si . This relation is valid, since Ass ≤ 0, s, s0 ∈ S, for any hyperbolic algebra [184].
For (U s , U s ) > 0 we get
εs > 0, (3.140)
s ∈ S. If θas > 0 for all s ∈ S, then (3.140) implies (see (??) and (2.44))
For pseudo-Euclidean metric g all ε(Is ) = 1 and, hence, all p-branes are Euclidean or should
contain even number of time directions: 2, 4, . . .. For ε[g] = 1 only magnetic p-branes may be
pseudo-Euclidean.
Example 5. F3 algebra. Now we consider an example of the solution corresponding to the
hyperbolic KM algebra F3 with the Cartan matrix
2 −2 0
A = −2 2 −1 , (3.143)
0 −1 2
CBPF-MO-003/02 38
F3 is an infinite dimensional Lie algebra generated by the (Serre) relations [186, 187]
for i = 1, 2, 3, respectively.
There exists an example of the solution with the A-matrix (3.149) for 11-dimensional model
governed by the action
Z p n
11 1 4 2 1 4∗ 2 o
S = d z |g| R[g] − (F ) − (F ) , (3.149)
4! 4!
where rankF 4 = rankF 4∗ = 4. Here ∆ = {4, 4∗}. We consider a configuration with two magnetic
5-branes corresponding to the form F 4 and one electric 2-brane corresponding to the form F 4∗ . We
denote S = {s1 , s2 , s3 }, as1 = as3 = 4, as2 = 4∗ and vs1 = vs3 = m, vs2 = e, where d(Is1 ) = d(Is3 ) = 6
and d(Is2 ) = 3. The intersection rules (3.122) read
where f = 5hts = H −5 > 0, h > 0 and ts > 0. The metric describes the power-law ”inflation”
2αi
in D = 11.PIt is singularP5for ts → 2+0. The powersi in scale-factors f do not satisfy Kasner-like
5
relations: i=1 di αi = i=1 di (αi ) = 1. For flat g the calculation of the Riemann tensor squared
gives us (see Appendix 1)
K[g] = 2, 1428t−4
s , (3.155)
CBPF-MO-003/02 39
where
K[g] ≡ RM N P Q [g]RM N P Q [g] (3.156)
is also called the Kretschmann scalar.
Example 6: H2 (q, q) algebra. Let
µ ¶
2 −q1
A= , q1 q2 > 4, (3.157)
−q2 2
q1 , q2 ∈ N. This is the Cartan matrix for the hyperbolic KM algebra H2 (q1 , q2 ) [186]. Let us
consider BD -model An example of the solution for BD -model with two electric p-branes (p = d1 , d2 ),
corresponding to F a and F b fields and intersecting in time manifold, is the following:
g = H −2/(q−2) ĝ 0 − H 2/(q−2) dt ⊗ dt + ĝ 1 + ĝ 2 , (3.158)
F = ν1 dH −1 ∧ dt ∧ τ1 + ν2 dH −1 ∧ dt ∧ τ2 , (3.159)
~ = −(~λa + ~λb )(q − 2)−1 ln H
ϕ (3.160)
where d0 = 3, d1 = a − 2, a = q + 4, b ≥ a, d2 = b − 2, d0 = 3, D = a + b. Here F = F a + F b for
a < b and F = F a for a = b. The signature restrictions are : ε1 = ε2 = −1. Thus, the space-time
(M, g) should contain at least three time directions. The minimal D is 14. For D = 14 we get
a = b = 7, d1 = d2 = 6, q = 3. In this case 5 ∩ 5 = 1.
3.1.2.3. Affine Lie algebras.
We note that affine KM algebras (with det A = 0) do not appear in the solutions (??)–(3.97).
Indeed, any affine Cartan matrix satisfy the relations
X
as0 As0 s = 0, (3.161)
s0 ∈S
with as > 0 called Coxeter labels [187], s ∈ S. This relation make impossible the existence of the
solution to eq. (3.89).
Thus, affine Cartan matrices do not arise in our solutions and hence some configurations are
(1)
forbidden. Let us consider A1 affine KM algebra with the Cartan matrix
µ ¶
2 −2
A= . (3.162)
−2 2
For D = 11 supergravity the intersections: 2 ∩ 5 = 0, 5 ∩ 5 = 2, corresponding to the A-matrix
(3.162), are forbidden.
Remark 6. In [193] new solutions in the affine case were obtained. These solutions contain as
a special case a solution in D = 11 supergravity from [133] with the intersection 5 ∩ 5 = 2. The
solutions from [193] use some modified ansatz for fields of forms (the ansatz for localized branes)
and do not belong to scheme under consideration. The solutions of this section in the special case of
D = 10, 11 supergravities are also different from the so-called non-marginal bound state solutions,
since the latter have non-trivial Chern-Simons terms (see, for example, [194, 195] and references
therein), although the rules for binary intersections may look similar.
where Xs is finite non-empty subset Xs ⊂ M0 , s ∈ S, all qsb > 0, and Xs = Xs0 , qsb = qs0 b for
0
b ∈ Xs = XsS 0 , s, s ∈ Sj , j = 1, . . . , k. The harmonic functions (3.163) are defined in domain
M0 \ X, X = s∈S Xs , and generate the solutions (??)–(3.97).
Denote S(b) ≡ {s ∈ S| b ∈ Xs }. We also put Mi = Rdi , R[g i ] = K[g i ] = 0, i = 1, . . . , n (see
definition (3.156)). Then for the metric (??) we obtain
C 0 + o(1)
K[g] = = [C + o(1)]|x − b|4(d0 −2)η(b) (3.164)
U 2 |x − b|4
for x → b ∈ X, where
X d(Is ) 1
η(b) ≡ (−εs )νs2 − , (3.165)
D − 2 d0 − 2
s∈S(b)
and C = C(b) ≥ 0 is given in Appendix 1 (see (??)). In what follows we consider non-exceptional
b ∈ X defined by relations C = C(b) > 0.
Remark 7. It follows from relation (??) of Appendix 1 that an exceptional point b ∈ X, defined
by relation C = C(b) = 0, appears iff (if and only if)
η(b) ≥ 0. (3.167)
From (3.165) we see that the metric (??) is regular at a “point” b ∈ X for εs = −1 and
large enough values of νs2 , s ∈ S(b). For εs = +1, s ∈ S(b), we have a curvature singularity at
non-exceptional point b ∈ X.
Now we consider a special case: d1 = 1, g 1 = −dt ⊗ dt. In this case we have a horizon, when
x → b ∈ X, iff
X 1
ξ1 (b) ≡ (−εs )νs2 δ1Is − ≥ 0. (3.168)
d0 − 2
s∈S(b)
This relation follows from the requirement of infinite time propagation of light to b ∈ X. If εs = −1,
1 ∈ Is for all s ∈ S(b), we get
η(b) < ξ1 (b), (3.169)
b ∈ X. This follows from the inequalities d(Is ) < D − 2 (d0 > 2).
We note that gtt → 0 for x → b ∈ X, if (3.169) is satisfied. This follows from the relation
x → b.
Remark 8. Due to relations (3.166) and (3.170) the point b ∈ X is non-exceptional if g 1 =
−dt ⊗ dt and 1 ∈ Is , εs = −1 for all s ∈ S(b).
Thus, for the metric (??) with Hs from (3.163) there are two dimensionless indicators at the
non-exceptional point b ∈ X: a) horizon indicator ξ1 (b) (corresponding to time t) and b) curva-
ture singularity indicator η(b). These indicators define (for our assumptions) the existence of a
horizon and the singularity of the Kretschmann scalar (when (Mi , g i ) are flat, i = 1, . . . , n) at
non-exceptional b ∈ X.
CBPF-MO-003/02 41
Generalized MP solutions. Here we consider special black hole solutions for the model
(7.1305) with all θa = 1, a ∈ ∆, when the signature of the metric gPis (−1, +1, . . . , +1). We
put εs = ε(Is ) = −1 and 1 ∈ Is for all s ∈ S and M0 = Rd0 , g 0 = dµ=1
0
dxµ ⊗ dxµ , M1 = R,
g 1 = −dt ⊗ dt. Then, the metric (??)-(3.94) reads
³Y 2
´1/(D−2) nX
d0
g= Hs2d(Is )νs dxµ ⊗ dxµ (3.171)
s∈S µ=1
³Y ´ n ³Y
X ´ o
−2νs2 −2νs2 δiIs
− Hs dt ⊗ dt + Hs ĝ i , (3.172)
s∈S i=2 s∈S
where (Mi , g i ) are flat Euclidean spaces, i = 2, . . . , n, Here all branes have a common time subman-
ifold M1 = R, for all a ∈ ∆, and
X d(Is ) 1
η(b) = νs2 − ≥ 0, (3.173)
D − 2 d0 − 2
s∈S(b)
b ∈ X. This solution describes a set of extreme p-brane black holes with horizons at b ∈ X. The
Riemann tensor squared has a finite limit at any b ∈ X.
Calculation of the Hawking ”temperature” corresponding to b ∈ X using standard formula (see,
for example, [399, 155]) gives us
TH (b) = 0, (3.174)
for any b ∈ X satisfying ξ1 (b) > 0.
Example 7: MP solution. The standard 4-dimensional Majumdar-Papapetrou solution [166]
in our notations reads
g = H 2 ĝ 0 − H −2 dt ⊗Fdt,
= νdH −1 ∧ dt, (3.175)
P
where ν 2 = 2, g 0 = 3i=1 dxi ⊗ dxi and H is a harmonic function. We have one electric 0-brane
(point) “attached” to the time manifold; d(Is ) = 1, εs = −1 and (U s , U s ) = 1/2. In this case (e.g.
for extremal Reissner-Nordström black hole) we get
and TH (b) = 0, b ∈ X.
Example 8: Solutions in D = 11 supergravity. In this case there are a lot of p-brane
MP-type solutions with orthogonal intersection rules [125]-[133], e.g. (i) solution with one electric
2-brane (d(Is ) = 3) and d0 = 8; (ii) solution with one magnetic 5-brane (d(Is ) = 6) and d0 = 5;
(iii) solution with one electric 2-brane and one magnetic 5-brane (d(Is1 ∩ Is2 ) = 2) and d0 = 4; (iv)
solution with two electric 2-branes (d(Is1 ∩ Is2 ) = 1) and d0 = 5; (v) solution with two magnetic
5-branes (d(Is1 ∩ Is2 ) = 4) and d0 = 3. In the examples (iii)-(v) the harmonic functions Hs1 and
Hs2 from (3.163) should have the coinciding sets of poles, i.e. Xs1 = Xs2 , to maintain the relation
(3.173). The Chern-Simons terms are zero for these solutions. In all these examples η(b) = 0,
b ∈ Xs and νs2 = 1/2, s ∈ S.
Fundamental matrix. Let
(na − 1)(nb − 1)
N (a, b) ≡ − λa · λb , (3.177)
D−2
a, b ∈ ∆. The matrix (3.177) is called the fundamental matrix of the model (7.1305) [179]. It
depends only on basic parameters of the model (7.1305), i.e. ranks of forms, total dimensions and
CBPF-MO-003/02 42
∆(s1 , s2 ) = D̄χ̄s1 χ̄s2 + n̄as1 χs1 χ̄s2 + n̄as2 χs2 χ̄s1 + N (as1 , as2 )χs1 χs2 , (3.178)
Ks = (U s , U s ) = K(as ), (3.184)
s ∈ S.
In most models including D = 11 supergravity, B12 theory [134], D < 11 supergravities [121],
K(a) = 2 and (3.122) has the following form
s1 6= s2 , and get As1 s2 = As2 s1 , i.e. the Cartan matrix is symmetric. In a finite dimensional case
we are led to the so-called simply laced or A − D − E Lie algebras. The intersection rules are
totally defined by the corresponding Dynkin diagram: d(Is1 ∩ Is2 ) = ∆(s1 , s2 ) − 1, when the vertices
corresponding to s1 and s2 are connected by a line and d(Is1 ∩ Is2 ) = ∆(s1 , s2 ) otherwise (since in
A − D − E case As1 s2 = 0, −1, s1 6= s2 ).
c = 1, . . . , k; s ∈ S.
We also consider the action
Z p
1
Sσ,0 = 2 dd0 x |g 0 |{R[g 0 ] − ĜAB g 0µν ∂µ σ̂ A ∂ν σ̂ B − 2V (σ̂)}, (3.188)
2κ0 M0
i.e. the action (2.36) with Φs = 0, s ∈ S, and the action (2.36) with omitted curvature and potential
terms
Z p n X o
1 s A
Sσ,1 = 2 |g 0 | −ĜAB g 0µν ∂µ σ̄ A ∂ν σ̄ B − εs e2UA σ̄ g 0µν ∂µ Φs ∂ν Φs . (3.189)
2κ0 M0 s∈S
Proposition 2. Let us consider the action (2.36), with the potential (3.186) satisfying orthog-
onality relations (3.187). Let metric g 0 and σ̂ = (σ̂ A (x)) satisfy equations of motion for the action
(3.188) and the constraints imposed:
UAs σ̂ A = 0, s ∈ S. (3.190)
Let g 0 , σ̄ = (σ̄ A (x)) and Φ = (Φs (x)) satisfy the equations of motion for the action (3.189) and
X
σ̄ A = U sA fs , (3.191)
s∈S
where fs = fs (x) are some functions, U sA = ĜAB UBs , (ĜAB ) = (ĜAB )−1 , s ∈ S. Then, the field
configuration
g 0 , σ = σ̂ + σ̄, Φ (3.192)
satisfies the equations of motion (??)–(2.47).
Proof. The proposition can be readily verified using the relations
ĜAB ∂µ σ̄ A ∂ν σ̂ B = 0; (3.193)
∂ ∂
V (σ̂ + σ̄) = V (σ̂), A
V (σ̂ + σ̄) = V (σ̂) (3.194)
∂σ ∂σ A
following from the conditions of Proposition 2.
Thus, we may find the exact solutions by two steps. First, we should solve the equations of
motion for the ”truncated” model (3.188) and find ”background” (σ̂, g 0 ) satisfying (3.190). On the
second stage we should solve the equations of motions corresponding to (3.189) for the fields σ̄ and
Φ on (M0 , g 0 )-background with the restriction of vanishing of total energy-momentum tensor for
(σ̄, Φ)-fields.
3.1.4.2. Generalized intersecting p-brane solutions with non-Ricci-flat spaces
Here consider the model (7.1305) with Λ = 0. Let the manifold be
instead of (7.1415). All (Mi , g i ) are Einstein spaces, satisfying (7.1313), with i = 1, . . . , n + k and
Then for electro-magnetic p-brane ansatz from Subsection 2.2 we get according to Proposition
0 the σ-model (2.36) with ”midisupermetric” (2.38); S, εs and UAs are defined in (2.25), (2.42), and
(2.40), respectively, and i, j = 1, . . . , n + k.
The potential V (σ), σ = (φi , ϕα ), in this case has the form (3.186) with
2di
uci = −2δin+c + , ucα = 0, (3.198)
2 − d0
Ac = −ξn+c dn+c /2, and c = 1, . . . , k; i = 1, . . . , n + k; α = 1, . . . , l.
It may be verified that the vectors uc = (ucA ) from (3.198) satisfy the orthogonality condition
(3.187). Indeed, the calculation gives for s = (a, v, I) (v = e, m and I ∈ Ωa,v )
d({n + c} ∩ I)
(uc , U s ) = 2 = 0, (3.199)
dn+c
since {n + c} ∩ I = ∅ for c = 1, . . . , k and I ∈ {1, . . . , n}.
Here Ω = Ω(n) is unchanged, so all p-branes do not ”live” in non-Ricci-flat ”internal” spaces
(Mn+c , g n+c ), c = 1, . . . , k.
Then from Propositions 1, 2 we obtain new exact solutions with the metric
( n+k
)
X i
g = U e2γ̂(x) ĝ 0 + Ui e2φ̂ (x) ĝ i (3.200)
i=1
n+k
1 X
γ̂ = γ0 (φ̂) = di φ̂i . (3.202)
2 − d0 i=1
The background fields g 0 and (σ̂ A ) = (φ̂i (x), ϕ̂α (x)) satisfy the equations of motion for the
σ-model (3.188) with (ĜAB ) defined in (2.38) and (Gij ) in (2.39), i, j = 1, . . . , n + k, and
n+k
1 X i
V (σ̂) = − ξi di e−2φ̂ +2γ0 (φ̂) . (3.203)
2 i=n+1
s ∈ S.
Relations (3.200), (3.201), (??) are the only modifications of the solutions (??)-(3.97). (All other
relations for F a , νs , . . . are unchanged).
CBPF-MO-003/02 45
on minisuperspace
M = RN , N = n + l + |S| (3.208)
(|S| is the number of elements in S) is defined by the relation
Gij 0 0
(GÂB̂ (X)) = 0 hαβ 0 . (3.209)
0 0 εs exp(−2U s (σ))δss0
Here we consider exact solutions to field equations corresponding to the action (3.206)
2V 0
Rµν [g 0 ] = GÂB̂ (X)∂µ X Â ∂ν X B̂ + g , (3.210)
d0 − 2 µν
1 p 1
p ∂µ [ |g 0 |GĈ B̂ (X)g 0µν ∂ν X B̂ ] − GÂB̂,Ĉ (X)g 0,µν ∂µ X Â ∂ν X B̂ = V,Ĉ , (3.211)
|g 0 | 2
the matrix (ĜAB ) is defined in (2.38). The zero-energy constraint (3.216) reads
1
EQ = ĜAB σ̇ A σ̇ B + VQ = 0. (3.220)
2
with
1
As = εs Q2s ,
Bs = 2Ks As , (3.224)
2
s ∈ S. These equations correspond to the Lagrangian
1 X X X
hs Ass0 q˙s q˙s0 −
0
LT L = As exp( Ass0 q s ), (3.225)
4 s,s0 ∈S s∈S s0 ∈S
s ∈ S.
Using (3.222) and (2.70) we obtain
X µ ¶
i d(Is )
φ = hs δiIs − q s + ci u + c̄i , (3.228)
s∈S
D−2
and X
ϕα = − hs χs λαas q s + cα u + c̄α , (3.229)
s∈S
and n n
0 1 X j 0 1 X j
c = dj c , c̄ = dj c̄ . (3.231)
2 − d0 j=1 2 − d0 j=1
where
1 X X X
hs Ass0 q˙s q˙s0 +
0
ET L = As exp( Ass0 q s ), (3.233)
4 s,s0 ∈S s∈S s0 ∈S
following from (3.121), (3.222), (??) and (3.227) we get for electric-type forms (??)
à !
Y −A 0
s
F = Qs fs0 ss dH ∧ τ (Is ), (3.235)
s0 ∈S
s ∈ Sm , where Q̄s = Qs ε(Is )µ(Is ), µ(I) = ±1 is defined in (2.58) and ∗0 = ∗[g 0 ] is the Hodge
operator on (M0 , g 0 ).
Relations for the metric and scalar fields follows from (??)-(3.230)
³Y ´n
g= fs2d(Is )hs /(D−2) exp(2c0 H + 2c̄0 )ĝ 0 (3.237)
s∈S
n ³Y
X ´ o
+ fs−2hs δiIs exp(2ci H + 2c̄i )ĝ i , (3.238)
i=1 s∈S
à !
Y hs χs λα
exp(ϕα ) = fs as
exp(cα H + c̄α ), (3.239)
s∈S
α = 1, . . . , l. Here
fs = fs (H) = exp(−q s (H)), (3.240)
where q s (u) is a solution to Toda-like equations (3.223) and H = H(x) (x ∈ M0 ) is a harmonic
function on (M0 , g 0 ).
The solution is presented by relations (3.235)-(3.240) with the functions q s defined in (3.223) and
the relations on the parameters of solutions cA , c̄A (A = i, α, 0), imposed in (??),(3.227), (3.231).
is the Cartan matrix of the Lie algebra Am and Bs > 0, s, s0 = 1, . . . , m. Here we put S =
{1, . . . , m}.
The equations of motion (3.223) correspond to the Lagrangian
m m
à m !
1 X 0
X X 0
LT = Ass0 q̇ s q̇ s − Bs exp Ass0 q s . (3.242)
2 s,s0 =1 s=1 s0 =1
CBPF-MO-003/02 49
This Lagrangian may be obtained from the standard one [226] by separating a coordinate describing
the motion of the center of mass.
Using the result of A. Anderson [227] we present the solution to eqs. (3.223) in the following
form
m+1
X
s
Cs exp(−q (u)) = vr1 · · · vrs ∆2 (wr1 , . . . , wrs ) exp[(wr1 + . . . + wrs )u], 3.2.B.3 (3.243)
r1 <...<rs
s = 1, . . . , m, where
s
Y ¡ ¢
∆(wr1 , . . . , wrs ) = wri − wrj ; ∆(wr1 ) ≡ 1, 3.2.B.4a (3.244)
i<j
denotes the Vandermonde determinant. The real constants vr and wr , r = 1, . . . , m + 1, obey the
relations
m+1
Y m+1
X
−2
vr = ∆ (w1 , . . . , wm+1 ), wr = 0.3.2.B.5 (3.245)
r=1 r=1
In (??)
m
Y ss0
Cs = Bs−A
0 , 3.2.B.6 (3.246)
s0 =1
0
where (Ass ) = (Ass0 )−1 is presented in (12.1970) of Appendix 3. Here
vr 6= 0, wr =
6 wr0 , r 6= r0 , 3.2.B.8 (3.247)
r, r0 = 1, . . . , m + 1. We note that the solution with Bs > 0 may be obtained ¿from the solution
with Bs = 1 (see [227]) by a certain shift q s 7→ q s + δ s .
The energy reads [227]
m m
à m ! m+1
1 X 0
X X 0 1X 2
ET = Ass0 q̇ s q̇ s + Bs exp Ass0 q s = wr .3.2.B.9 (3.248)
2 s,s0 =1 s=1 0
s =1
2 r=1
Here Ks = K, hs = h = K −1 , s ∈ S.
Thus, in the Am Toda chain case eqs. (??)-(??) should be substituted into relations (3.233) and
(3.235)-(3.240) for the general solution.
CBPF-MO-003/02 50
Examples for d0 > 2. Here we consider the case d0 > 2. Let matrix (hαβ ) be positively defined
and K = Ks > 0. Then from the energy constraint (3.233) we get
ET L ≤ 0 =⇒ ET ≤ 0. (3.250)
In this case
cα = ci = 0 ⇐⇒ ET L = ET = 0, (3.251)
i = 1, . . . , n, and α = 1, . . . , l. When (hαβ ) is negative definite (as it takes place in 12-dimensional
theory from [134]) there exist solutions with ET L > 0.
A1 -case. Here we consider the case of one “brane”, i.e. S = {s}. Solving the Liouville equation
we get
fs (H) = |Ks |1/2 |Qs |fˆs (H), 3.2.B.15a (3.253)
where H = H(x) > 0 and
1 p
fˆs (H) = √ ch( ET H), εs Ks > 0, ET > 0; (3.254)
ET
1 p
√ sh( ET H), εs Ks < 0, ET > 0; (3.255)
ET
1 p
√ sin( −ET H), εs Ks < 0, ET < 0; (3.256)
−ET
H, εs Ks < 0, ET = 0. (3.257)
Here
ET = (q̇ s )2 + Bs exp(2q s ).3.2.B.9b (3.258)
In a special case ET = 0 this solution agrees with those from [147, 163] if the following redefinition
of the harmonic function is performed: H 7→ |Ks |1/2 |Qs |H.
A2 -case. Now we consider the case m = 2. We put S = {1, 2}. The solution reads
where
w1 + w2 + w3 = 0, (3.261)
v1 v2 v3 = (w1 − w2 ) (w2 − w3 )−2 (w3 − w1 )−2 .
−2
(3.262)
and
C1 = (B12 B2 )−1/3 , C2 = (B22 B1 )−1/3 .3.2.C.5 (3.263)
Let K > 0. Then ET ≤ 0 and hence some of Bi should be negative.
Let B1 < 0 and B2 < 0. In the pseudo-Euclidean case, when ε[g] = −1 and all θa = 1, this
means that εsi = ε(Isi ) = −1, i = 1, 2, i.e. all p-branes should contain an odd number of time
submanifolds.
CBPF-MO-003/02 51
where |C1 | = (B12 |B2 |)−1/3 , |C2 | = (B22 |B1 |)−1/3 , |Bj | = Q2j , j = 1, 2, and
of motion of such a model are reduced to the Euler-Lagrange equations for the so-called pseudo-
Euclidean Toda-like Lagrange system. We reproduce this result in 3.3.2. Methods for integrating of
pseudo-Euclidean Toda-like systems (see, [ 73] and references therein) are based on a Minkowski-like
geometry for the characteristic vectors determining the potential of the pseudo-Euclidean Toda-like
system. If the characteristic vectors form an orthogonal set, then the pseudo-Euclidean Toda-like
system is integrable. The corresponding p-brane models have been studied in [179], [156], [214],
[3]. Here we apply these methods for integrating the p-brane model reducible to the algebraic gen-
eralizations of an open Toda chain. The characteristic vectors of such models may be interpreted
as root vectors of the semi-simple Lie algebra. We consider the Lie algebras of the Cartan types
Ar , Br , Cr . Using the technique suggested by Anderson [227] for solving the Toda chain’s equations
of motion, in 3.3.3, 3.3.4 we integrate the p-brane models. In the last subsection we examine the
metric obtained for some particular exact solution appearing for a quite wide class of p-brane mod-
els. This solution describes the nonextremal black hole under some condition. The corresponding
ADM-mass and the Hawking temperature of a black hole are calculated.
where ds2 = gM N dz M dz N is the metric with Lorentzian signature on the manifold M (M, N =
0, 1 . . . , D − 1), |g| = | det(gM N )|. (hαβ ) is a symmetrical positively definite ω × ω matrix, λa (ϕ) is
a linear combination of the scalar fields, i.e.
ω
X
λa (ϕ) = λa,α ϕα , (3.271)
α=1
The field AaM1 ,...,Mna −1 may be called a gauge potential corresponding to the field strength FM
a
1 ,...,Mna
.
By ∆ we denote some finite set.
The action (3.195) leads to the following equations of motion
1
RM N − gM N R = TM N , (3.275)
2
X 1
α
4[g]ϕ = λαa e2λa (ϕ) (F a )2 , α = 1, . . . , ω, (3.276)
a∈∆
na !
5M1 [g]( e2λa (ϕ) F a,M1 ...Mna ) = 0, a ∈ ∆. (3.277)
CBPF-MO-003/02 53
TM N = TM N [ϕ] + TM N [F ], (3.278)
where we denoted
ω
X µ ¶
1 α α P β β
TM N [ϕ] = hαβ ∂M ϕ ∂N ϕ − gM N ∂P ϕ ∂ ϕ , (3.279)
α,β=1
2
X e2λa (ϕ) µ 1 a,M2 ...Mna
¶
a 2 a
TM N [F ] = − gM N (F ) + na FM M2 ...Mna FN . (3.280)
a∈∆
na ! 2
In (3.276),(3.277) we denoted the Laplace-Beltrami operator and covariant derivative with respect
to the metric gM N by 4[g] and 5M [g], respectively. The constants λαa in (3.276) are introduced by
ω
X
λαa = hαβ λa,β , (3.281)
α,β=1
M0 ≡ S d0 , (3.284)
M1 ≡ Rt , d1 = 1. (3.285)
Clearly, the canonical projection p̂i : M → Mi of τi provides with the volume form of Mi .
CBPF-MO-003/02 54
In order to construct the p-brane worldvolumes we introduce submanifolds of the following type
where
I = {i1 , . . . , ir }, i1 < . . . < ir , (3.289)
is any ordered non-empty subset of natural numbers 2, . . . , n. Let Ω0 be the set of all such elements
including the empty set, i.e.
By definition, put
I¯ ≡ {2, . . . , n} \ I. (3.291)
In this paper we consider electrically charged p-branes with the following worldvolumes
(e)
MI = Rt × MI , I = {i1 , . . . , ir } ∈ Ω0 . (3.292)
(e) (e)
For empty I = ∅ we put MI = Rt . The dimension of MI is given by
(e)
d(I) ≡ dim MI = 1 + di1 + . . . + dir . (3.293)
(e)
(d(I) = 1 for I = ∅). The canonical projection p̂I : M → MI of the following d(I)-form
where Φ(a,e,I) (u) is a scalar function, describes an electrically charged p-brane (p = na − 2) with the
(e)
worldvolume MI . Moreover, the submanifold R+ × S d0 × MI¯ (R+ × S d0 for I = {2, . . . , n}) is the
so-called transverse space for this p-brane. The na -form field strength corresponding to A(a,e,I) was
defined by (3.197) and may be written as
In this paper we consider the so-called composite p-branes [152], i.e., by definition we put
X X
Fa = F (a,e,I) + F (a,m,J) , (3.300)
I∈Ωa,e J∈Ωa,m
where Ωa,e ⊂ Ω0 is a subset (which may be empty) of all I ∈ Ω0 such that d(I) + 1 = na ≡
rank F (a,e,I) . Moreover, Ωa,m ⊂ Ω0 is a subset (which may be empty) of all J ∈ Ω0 such that
(m)
D − d(J) − 1 = dim MJ = na ≡ rank F (a,m,J) . Evidently, Ωa,m = ∅ for na = D − 1, D.
We obtain the following non-zero components of the Ricci tensor for the metric (3.208)
à n !
X
0 −2γ k 2
R0 = − e dk (ẋ ) + γ¨0 − γ̇ γ˙0 , (3.301)
k=0
n k £ ¤ o
Rnmkk = δ0k (d0 − 1) e−2x − ẍk + ẋk (γ˙0 − γ̇) e−2γ δnmkk , (3.302)
where we denoted n
X
γ0 = dk xk . (3.303)
k=0
Pn Pn
Indices mk and nk in (7.1329)
Pn for k = 0, . . . , n run over from (D − l=k dl ) to (D − l=k dl + dk − 1).
We recall that D = 1 + k=0 dk = dim M.
Under the above assumptions related to the F a -fields and the metric (3.208) the Maxwell-like
equations (3.277) and the Bianchi identities dF a = 0 have the following form, correspondingly
d h γ0 −γ−2σ(I)+2λa (φ) (a,e,I) i
e Φ̇ (u) = 0, I ∈ Ωa,e , a ∈ ∆ (3.304)
du
d h γ0 −γ−2σ(J)−2λa (φ) (a,m,J) i
e Φ̇ (u) = 0, J ∈ Ωa,m , a ∈ ∆, (3.305)
du
where X
σ(I) = d1 x1 + di xi . (3.306)
i∈I
where
χs = +1, vs = e, (3.310)
χs = −1, vs = m. (3.311)
CBPF-MO-003/02 56
where i, j = 2, . . . , n. Here the intersection I ∩J may be empty. The total energy-momentum tensor
of F a -fields has a block-diagonal form whenever the restriction (3.237) is valid. Using (3.234), we
present its non-zero components in the form
1 X
T [F a ]00 = − e−2γ0 Q2s exp [2σ(Is ) − 2χs λas (φ)] , (3.313)
2 s∈S∗
à !
1 −2γ0 X
T [F a ]m
nk
k
=− e (2δkIs − 1)Q2s exp [2σ(Is ) − 2χs λas (φ)] δnmkk , (3.314)
2 s∈S ∗
where X
δkI = δki + δk1 , I ∈ Ω0 , k = 0, 1, . . . , n. (3.315)
i∈I
We put δkI = δk1 for I = ∅. Evidently, δ0I = 0 and δ1I = 1 for any I ∈ Ω0 .
We assume that the dilatonic scalar fields ϕα depend only on the radial coordinate u. Under
this assumption the total energy-momentum tensor of the dilatonic scalar fields reads
à ω !
¡ ¢ 1 X
T [ϕ]M
N = e−2γ hαβ ϕ̇α ϕ̇β diag(1, 1, . . . , 1, −1, 1, . . . , 1). (3.316)
2 | {z }
α,β=1 d0 times
M
The Einstein equations (3.200) can be written as RN = TNM − T δN
M
/(D − 2). Further we employ
0 0 mk mk mk
the equations R0 − R/2 = T0 and Rnk = Tnk − T δnk /(D − 2). Using (3.226), (3.227), (3.238),
(3.239), (3.241), we obtain these equations in the form
à n ω
!
1 X X
Gkl ẋk ẋl + hαβ ϕ̇α ϕ̇β + V = 0, (3.317)
2 k,l=0 α,β=1
½
k k 2γ k
ẍ + (γ˙0 − γ̇)ẋ = e δ0k (d0 − 1) e−2x
X µ d(Is )
¶ ¾
−2γ0 2
+ e χs δkIs − Qs exp [2σ(Is ) − 2χs λas (φ)] (3.318)
s∈S
D − 2
∗
where we denoted
Gkl = dk δkl − dk dl , k, l = 0, 1, . . . , n, (3.319)
( )
1 X
−2x0
V = e2γ (d0 − 1)d0 e − e−2γ0 Q2s exp [2σ(Is ) − 2χs λas (φ)] . (3.320)
2 s∈S∗
CBPF-MO-003/02 57
It is not difficult to verify that equations (3.242),(3.243),(3.246) may be presented as the Euler-
Lagrange equations obtained from the Lagrangian
" Ã n ω
! #
1 X X
L = eγ0 −γ Gkl ẋk ẋl + hαβ ϕ̇α ϕ̇β − V (3.322)
2 k,l=0 α,β=1
It is easy to check that the radial coordinate u is a harmonic function in this gauge, i.e. 4[g]u = 0.
Let us introduce an (n + ω + 1)-dimensional real vector space Rn+ω+1 . Denote by eA , A =
0, 1, . . . , n + ω, the canonical basis in Rn+ω+1 (e1 = (1, 0, . . . , 0) etc.). Define the following vectors:
1. The vector whose coordinates are to be found
n+ω
X ¡ ¢
x(u) = xA (u)eA , xA (u) = (x0 (u), . . . , xn (u), ϕ1 (u), . . . , ϕω (u)). (3.324)
A=0
2. The vector corresponding to the factor-space M0 ≡ S d0 with a non-zero Ricci tensor. Here-
after, we call it by the vector induced by the curvature of M0
n+ω
X µ ¶
A 2 ¡ A¢ 1
V0 = V0 eA = − e0 , V0 = −2 , 0, . . . , 0 . (3.325)
A=0
d0 d0
A set of the vectors Us characterizes the space-time M, the configuration of p-branes and their
couplings to dilatonic scalar fields. Further these vectors are called characteristic.
Let < ., . > be a symmetrical bilinear form on Rn+ω+1 such that
< eA , eB >= ḠAB , (3.327)
where we put by definition
µ ¶
Gkl 0
(ḠAB ) = . (3.328)
0 hαβ
CBPF-MO-003/02 58
The form < ., . > endows the space Rn+ω+1 with the metric, whose signature is (−, +, . . . , +) [179].
By the usual way we may introduce the covariant components of vectors. For the vectors V0 , Us the
covariant components have the form
The values of the bilinear form < ., . > for V0 , Us look as follows
d0 − 1
< V0 , V0 >= −4 , (3.332)
d0
< V0 , Us >= 0, ∀s ∈ S, (3.333)
" ω
#
d(Is )d(Is0 ) X
< Us , Us0 >= 4 d(Is ∩ Is0 ) − + χs χs0 hαβ λαas λβas0 , (3.334)
D−2 α,β=1
where x, bµ ∈ Rn+ω+1 . It should be noted that the kinetic term < ẋ, ẋ > is not a positively definite
quadratic form as there usually takes place in classical mechanics. Due to the pseudo-Euclidean
signature (−, +, ..., +) of the form < ., . > such systems may be called the pseudo-Euclidean Toda-
like systems as the potential like that given in (3.264) defines the algebraic generalizations of the
Toda chain [226]. well-known in classical mechanics .
3.3.3 Integration of the p-brane model with linearly independent characteristic vectors
We recall that S∗ ⊂ S is the subset of all s ∈ S such that Qs 6= 0. Define a bijection f : S∗ 7→
{1, 2, . . . , r}, where we denote by r the cardinal number of S∗ , i.e.
r = |S∗ |. (3.340)
Denote the natural number f (s) corresponding to s ∈ S∗ by the same letter s.
The problem consists in integrating the equations of motion obtained from the Lagrangian
(3.260) under the zero-energy constraint (3.336). Suppose the characteristic vectors Us ∈ Rn+ω+1 ,
induced by p-branes are linearly independent. Then r ≤ n + ω. We introduce a basis {fA } in
Rn+ω+1 in the following manner
V0 2Us
f0 = , fs = , s = 1, . . . , r, (3.341)
< V 0 , V0 > < U s , Us >
< fA , fp >= δAp , A = 0, . . . , n + ω, p = r + 1, . . . , n + ω. (3.342)
Notice that if r ≡ |S∗ | = n + ω then the basis {fA } does not contain the vectors fp with p ≥ r + 1.
We also mention that due to the relation (3.333) we get < f0 , fs >= 0 for s = 1, . . . , r. It is not
difficult to prove that the vectors f1 , . . . , fr , fr+1 , . . . , fn+ω must be space-like.
Using the decomposition
r
X n+ω
X
x(u) = q 0 (u)f0 + [q s (u) − ln C s ]fs + q p (u)fp , (3.343)
s=1 p=r+1
we present the Lagrangian (3.260) and the constraint (3.261) in the form
L = L0 + LT + LP , (3.344)
E = E0 + ET + EP ≡ 0, (3.345)
where
(q̇ 0 )2 d0 (d0 − 1) q0
L0 = − e , (3.346)
2 < V0 , V 0 > 2
(q̇ 0 )2 d0 (d0 − 1) q0
E0 = + e , (3.347)
2 < V0 , V 0 > 2
r r
" r #
X Css0 0
X 2 X 0
LT = q̇ s q̇ s + exp Css0 q s , (3.348)
s,s0 =1
< U s , U s > s=1
< U s , U s > s0 =1
r r
" r
#
X Css 0 0
X 2 X 0
ET = q̇ s q̇ s − exp Css0 q s , (3.349)
s,s0 =1
< U s , Us > s=1
< U s , Us > s0 =1
n+ω
1 X p 2
LP = EP = (q̇ ) . (3.350)
2 p=r+1
CBPF-MO-003/02 60
We introduced, in (3.273), (3.274) the nondegenerate Cartan-type matrix (Css0 ) by the following
manner
< U s , U s0 >
Css0 = 2 , s, s0 = 1, . . . , r. (3.351)
< Us0 , Us0 >
The constants C s in the decomposition (3.268) are defined by
r · ¸C ss 0
Y < Us0 , Us0 > 2
s
C = Qs0 , s = 1, . . . , r, (3.352)
0
s =1
4
0
where (C ss ) is the inverse matrix to (Css0 ).
The Euler-Lagrange equations for q r+1 (u), . . . , q n+ω (u) read q̈ r+1 (u) = . . . = q̈ n+ω (u) = 0. Inte-
grating them, we get
q p (u) = ap u + bp , p = r + 1, . . . , n + ω, (3.353)
n+ω
1 X p 2
EP = (a ) ≥ 0, (3.354)
2 p=r+1
The set of equations (3.285) proved to be completely integrable if (Css0 ) is the Cartan matrix of
a simple complex Lie algebra. The general solutions for some algebras as well as some particular
solution of the set (3.285) for quite a wide class of matrices (Css0 ) will be considered in the next
sections. Here we suppose that the functions are known and the corresponding integral of motion
(3.274) is calculated.
Combining (3.266),(3.278),(3.280)(3.286), we present the decomposition (3.268) in the following
form
X r
2V0 2Us
x(u) = − ln[F0 (u − u0 )] − ln[C s Fs (u)] + uQ + P, (3.363)
< V0 , V0 > s=1 < U s , Us >
Due to the assumptions (3.267) their coordinates QA , P A w.r.t. the canonical basis {eA }satisfy the
constraints
n
X n
X
k
< Q, V0 >= 2 Q (dk − δk0 ) = 0, < P, V0 >= 2 P k (dk − δk0 ) = 0. (3.365)
k=0 k=0
n+ω
X n+ω
X
A
< Q, Us >= Q Us,A = 0, < P, Us >= P A Us,A = 0, s = 1, . . . , r. (3.366)
A=0 A=0
for the p-brane with magnetic-type charge. We put F (as ,m,Is ) = Qs τ0 if I¯s = ∅. We stress that if
r ≡ |S∗ | = n + ω, then one must put QA = P A = 0, A = 0, . . . , n + ω in this solution.
where ws are arbitrary constants, which may be real (including zero) or imaginary.
2. (Css0 ) = (2δss0 −δs,s0 +1 −δs,s0 −1 ) is the Cartan matrix of the simple Lie algebra Ar ≡ sl(r+1, C).
In this case all characteristic vectors Us are space-like with coinciding lengths, i.e.
By the transformation
πi
q s 7→ q s − ms , (3.376)
2
where r
X 0
ms = 2 C ss = s(r + 1 − s), (3.377)
s0 =1
These are precisely the Ar Toda equations [226]. Using the general solutions to these equations
presented by Anderson [227], we obtain the following result
r+1
X
s(r+1−s)
Fs (u) = i vµ1 · · · vµs ∆2 (µ1 , . . . , µs ) e(wµ1 +...+wµs )u , (3.379)
µ1 <...<µs
r+1
1X 2
ET = w . (3.380)
2 µ=1 µ
CBPF-MO-003/02 63
The constants vµ , wµ are in general complex. There are additional constraints on them if one
requires the functions Fs (u) and the integral of motion (3.305) to be real. In (3.304) we used
µs = s(r + 1 − s) for Ar .
The Cartan matrix of the simple Lie algebra Br ≡ so(2r + 1) may be obtained from that given
in (3.308) by multiplying the last column of (Csr ) by 2. In this case the general solution to
the set of equations ( 3.285) may be obtained from the previous formulae (3.304), (3.305) as
in [332]. Notice that the equations (3.285) are symmetric under the following permutations
q s ↔ q r+1−s for s = 1, . . . , r if (Css0 ) if the Cartan matrix of Ar . This implies that there
are solutions (3.304) with q s ≡ q r+1−s for s = 1, . . . , r. Moreover, this identification for
r = 4, 6, 8, . . . leads to the (r/2) equations of the form (3.285) with the matrix (3.308).
Consequently the general solution of the equations (3.285) with the matrix (3.308) for r = r0
may be obtained form (3.304) for r = 2r0 by putting additional constraints on the constants
vµ , wµ providing with the identities Fs (u) ≡ F2r0 +1−s (u), s = 1, . . . , 2r0 .
4. (Css0 ) is the Cartan matrix of the simple Lie algebra Cr ≡ sp(r, C), i.e.
½
2δss0 − δs−1,s0 − δs−1,s0 for s = 1, . . . , r − 1, s0 = 1, . . . , r,
Css0 = (3.384)
δss0 − δs,s0 −1 for s = r, s0 = 1, . . . , r.
In this case the general solution to the set of equations (3.285) with r = r0 may be obtained
from (3.304) with r = 2r0 − 1 by putting additional constraints on the constants vµ , wµ
providing with identities Fs (u) ≡ F2r0 −s (u), s = 1, . . . , 2r0 − 1. It stems ¿from the following
property of the set (3.285): the identification q s ≡ q 2r0 −s for s = 1, . . . , 2r0 − 1 reduces the
set (3.285) with the Cartan matrix of A2r0 −1 to the set (3.285) with the Cartan matrix of Cr0
(r0 ≥ 2).
CBPF-MO-003/02 64
The conditions are valid for extremely large class of the p-brane models. For instance, the parameters
ms are natural numbers if (Css0 ) is the Cartan matrix of a semi-simple Lie algebra G [237]. For
G = Ar = sl(r + 1, C) the parameters ms are given by (3.302).
Under the conditions (3.310) the set of equations (3.287) admits the following particular solution
µ ¶m
sinh[µ̄(u − u01 )] s
Fs (u) = as , s = 1, . . . , r, (3.386)
µ̄
and the constants µ̄, u01 are arbitrary. The corresponding to (3.311) integral of motion (3.274) has
the form
r
X r r r
Css0 Ḟs Ḟs0 X 2 Y
−Css0 2
X ms
ET = − (Fs0 ) = 2µ̄ . (3.388)
0
s,s =1
< U s , Us > Fs Fs0 s=1
< U s , Us > 0
s =1 s=1
< U s , Us >
It is evident that (3.314) represents the polynomials in the radial coordinate u if the parameters
ms are natural numbers. As we have already mentioned, ms are natural numbers if, for instance,
(Css0 ) is the Cartan matrix of a semi-simple Lie algebra. The formula (3.314) does not exhaust
all possible polynomial solutions to the set (3.287). As far as we know, an explicit general form
of the polynomial solution, which appears for the set of equations (3.287) with arbitrary natural
numbers ms and vanishing integral of motion (3.313), is not found. There are only few examples in
the literature. For instance, in [329] all possible polynomial solutions were obtained for the matrix
(Css0 ) supposed to be the Cartan matrix of the Lie algebras Ar ≡ sl(r + 1, C) with r = 1, 2, 3. It is
easy to check that an arbitrary polynomial solution to (3.287) under the condition ET = 0 may be
obtained from (3.314) by adding some polynomial of lower degree to the leading term as ums .
Now we use the particular solution (3.311) with µ̄ > 0 and its special form (3.314) corresponding
to µ̄ = 0 for constructing non-extremal and extremal black holes, respectively. Consider the general
form of exact solution (3.292)-(3.296) under the following additional assumptions
1. We put
à r
!
2V0A X 2ms UsA
QA = µ̄ + − δ1A , A = 0, . . . , n + ω. (3.391)
< V0 , V0 > s=1 < Us , Us >
CBPF-MO-003/02 65
One may verify that the conditions < Q, V0 >= 0, < Q, Us >= 0, s = 1, . . . , r are valid.
Using the zero-energy constraint (3.270) and (3.313), we find the constant E0
1 1
E0 = −ET − < Q, Q >= − d0 d¯0 µ2 , (3.392)
2 2
where we denoted
µ = µ̄/d¯0 , d¯0 = d0 − 1. (3.393)
where ε0 is an arbitrary positive constant. The conditions < P, V0 >= 0 lead to the following
constraint on parameters R2 , . . . , Rn
n
X
Ri di = d¯0 . (3.395)
i=2
3. Now we consider the solution for u ∈ (u0 , +∞) and introduce the following new radial coor-
dinate
µ ¶1/d¯0
R0 2µ
R= = ε0 , R > R0 . (3.397)
1 − exp[−2µ̄(u − u0 )] 1 − exp[−2µ̄(u − u0 )]
The constant R0 is defined by
¯
R0 = ε0 (2µ)1/d0 . (3.398)
Here we take the constant u01 in (3.311) such that (u0 − u01 ) > 0. Moreover we introduce the
constant µ ¶1/d¯0
2µ
R∗ = R |u=2u0 −u01 = ε0 > R0 . (3.399)
1 − exp[−2µ̄(u0 − u01 )]
Finally, we obtain the metric
" µ ¶d¯0 µ ¶d¯0 #
Pr 8ms d(Is )
s=1 (D−2)<Us ,Us > ½
2 R∗ R0 dR2
ds = 1+ − + R2 dΩ2d0
R R 1 − (R0 /R)d¯0
" µ ¶d¯0 µ ¶d¯0 #−
Pr 8ms
s=1 <Us ,Us >
à µ ¶d¯0 !
R∗ R0 R0
− 1+ − 1− dt2
R R R
n
" µ ¶d¯0 µ ¶d¯0 #−
Pr 8ms δiI
s
¾
X R∗ R0
s=1 <Us ,Us >
−2Ri
+ ε0 1+ − ds2i , (3.400)
i=2
R R
CBPF-MO-003/02 66
Then constants µ̄, ε0 , (u0 − u01 ) are independent. The constants µ, d¯0 , R0 , R∗ are defined by
(3.318),(3.323), (3.324). The parameters R2 , . . . , Rn obey the relation (3.320).
Now we analyze the particular solution (3.325)-(3.327). The metric (3.325) is asymptotically
flat, i.e.
X n
2 2 2 2 2 i
lim ds = dR + R dΩd0 − dt + ε−2R
0 ds2i . (3.404)
R→+∞
i=2
According to (3.320) all parameters R2 , . . . , Rn may be positive. Then, the constant scale factors
i
ε−2R
0 of internal spaces M2 , . . . , Mn are arbitrary small if ε0 is large enough.
If µ̄ > 0 the particular solution describes a non-extremal black hole with the horizon at R = R0 .
The active gravitational mass Mg and the Hawking temperature TH of this black hole read
"Ã µ ¶d¯0 ! X r µ ¶d¯0 #
1
¯ R 0 4m U
s s R0
2GN Mg = R∗d0 1− + , (3.405)
R∗ s=1
< U s , U s > R ∗
µ ¶Pr 4ms
d¯0
−1
R0 s=1 <Us ,Us >
TH = , (3.406)
4πkB R∗ R∗
where GN and kb are Newton’s gravitational constant and Boltzmann’s constant, respectively.
The solution (3.325)-(3.327) may be considered in the so-called extreme case, when µ̄ = 0
(R0 = 0). It follows from general statements proved in [155] that the point R = 0 is a curvature
singularity of the metric (3.325) with R0 = 0 if TH → +∞ as R0 → +0. Then, the particular
solution admits an extremal black hole only if
r
X 4ms
≥ 1. (3.407)
s=1
< U s , Us >
CBPF-MO-003/02 67
where ẋ ≡ dx/du,
n
2γ0 (φ) wX i
Vw = −wV = −wΛ e + ξi di e−2φ +2γ0 (φ) (4.411)
2 i=1
is the potential with
n
X
γ0 (φ) ≡ di φi , (4.412)
i=1
and
N = exp(γ0 − γ) > 0 (4.413)
is the lapse function, U s = U s (φ, ϕ) are defined in (2.40), εs are defined in (2.42) for s = (as , vs , Is ) ∈
S, and
Gij = di δij − di dj (4.414)
are components of the ”pure cosmological” minisupermetric, i, j = 1, . . . , n [55].
In the electric case (F (a,m,I) = 0) for finite internal space volumes Vi the action (??) coincides
with the action (7.1305) if µ = −w/κ20 , κ2 = κ20 V1 . . . Vn .
Action (??) may be also written in the form
Z n o
µ
Sσ = duN GÂB̂ (X)Ẋ Â Ẋ B̂ − 2N −2 Vw , (4.415)
2
where σ = (σ A ) = (φi , ϕα ),
where
The integrability of the Lagrange system (??) crucially depends upon the scalar products of
co-vectors U Λ , U j , U s (see (3.82)). These products are defined by (3.85) and the following relations
[147]
δij
(U i , U j ) = − 1, (4.421)
dj
(U i , U Λ ) = −1, (4.422)
D−1
(U Λ , U Λ ) = − , (4.423)
D−2
(U s , U i ) = −δiIs , (4.424)
d(Is )
(U s , U Λ ) = , (4.425)
2−D
where s = (as , vs , Is ) ∈ S; i, j = 1, . . . , n.
Toda-like representation. We put γ = γ0 (φ), i.e. the harmonic time gauge is considered.
Integrating the Lagrange equations corresponding to Φs (see (3.217)) we are led to the Lagrangian
¿from (3.218) and the zero-energy constraint (3.220) with the modified potential
1X
VQ = Vw + εs Q2s exp[2U s (σ)], (4.426)
2 s∈S
n ³Y
X ´ o
+ fs−2hs δiIs i
exp(2c u + 2c̄ )ĝ , i i
i=1 s∈S
à !
Y hs χs λα
α
exp(ϕ ) = fs as
exp(cα u + c̄α ), (4.428)
s∈S
α = 1, . . . , l, where fs = fs (u) = exp(−q s (u)) and q s (u) obey Toda-like equations (3.223).
Relations (3.231) and (3.233) take the form
n
X n
X
0 j 0
c = dj c , c̄ = dj c̄j , (4.429)
j=1 j=1
n
à n !2
X X
2E = 2ET L + hαβ cα cβ + di (ci )2 − di ci = 0, (4.430)
i=1 i=1
with ET L from (3.233) and all other relations (e.g. constraints (??), (3.227)) and relations for forms
(3.235) and (3.236) with H = u) are unchanged. In a special Am Toda chain case this solution was
considered previously in [224].
ξ1 6= 0, ξ2 = . . . = ξn = 0, (4.431)
1∈
/ Is , (4.432)
U 1A X U sA
σ A (u) = − ln |f1 (u)| − ln(fs (u)) + cA u + c̄A , (4.435)
(U 1 , U 1 ) s∈S
(U s, U s)
CBPF-MO-003/02 70
where
p
f1 (u) = R sh( C1 (u − u1 )), C1 > 0, ξ1 w > 0; (4.436)
p
R sin( |C1 |(u − u1 )), C1 < 0, ξ1 w > 0; (4.437)
p
R ch( C1 (u − u1 )), C1 > 0, ξ1 w < 0; (4.438)
1/2
|ξ1 (d1 − 1)| , C1 = 0, ξ1 w > 0, (4.439)
Forms à !
Y −Ass0
F s = Qs fs 0 du ∧ τ (Is ), (4.447)
s0 ∈S
F s = Qs τ (I¯s ), (4.448)
correspond to magnetic p-branes; Qs 6= 0, s ∈ S. Remind that fs = fs (u) = exp(−q s (u)) with q s (u)
obeying Toda-like equations (3.223).
Restriction 1 ( see Subsect. 2.3.1) forbids certain intersections of two p-branes with the same
color index for n1 ≥ 2. Restriction 2 is satisfied identically in this case.
This solution in a special case of Am Toda chain was obtained earlier in [222] (see also [224]).
Some special configurations were considered earlier in [160, 135, 159].
CBPF-MO-003/02 71
i = 0, . . . , n, [209, 112, 74, 75] that was also generalized to some other matter fields, e.g scalar one
(see [212] and reference therein).
The solution is given by the relation (??) and (??)-(4.460) (see below). In the non-composite
case the ”cosmological derivation” of the solution was obtained in [165]. The Freund-Rubin solutions
[200] in D = 11 supergravity: AdS4 × S 7 and AdS7 × S 4 , correspond to M2-brane ”living” on AdS4
and S 4 respectively. The ”popular” AdS5 × S 5 solution in IIB (D = 10) supergravity model [204]
corresponds to composite self-dual configuration with two ”branes” living on AdS5 and S 5 and
corresponding to 5-form.
We consider the model governed by the action (7.1305). The equations of motion were presented
in (3.273)-(7.1410). The Hilbert-Einstein equations (3.273) may be written in the equivalent form
(2.63). Here we keep the multi-index notations with Ω = Ω0 being a set of all non-empty subsets of
{0, . . . , n}.
The solution reads as following: the manifold and the metric are defined by the relations (7.1310),
(??), (7.1313) (with i = 0, . . . , n), the fields of forms and scalar fields are the following
X
Fa = QaI τ (I), (4.457)
I∈Ωa
α
ϕ = const. (4.458)
Here Ωa ⊂ Ω are non-empty subsets satisfying the Restriction presented below and QaI are
constants, I ∈ Ωa , a ∈ ∆. The parameters of solution obey the relations
X X
θa λαa e2λa (ϕ) (QaI )2 ε(I) = 0, (4.459)
a∈∆ I∈Ωa
2Λ X X na − 1
ξi = + θa e2λa (ϕ) (QaI )2 ε(I)[δIi − ], (4.460)
D − 2 a∈∆ I∈Ω
D−2
a
i = 0, . . . , n.
The solution is valid if the following restriction on the sets Ωa , a ∈ ∆, (similar to electric part
of Restriction 1) is satisfied.
Restriction 3. For any a ∈ ∆ and I, J ∈ Ωa
d(I ∩ J) ≤ na − 2 (4.461)
Due to relations from the Appendix 2 (12.1948), (12.1949), (12.1956) and (12.1957) this restric-
tion guarantees the block-diagonal structure of the ZM N -tensor in (2.65).
From (??) the self-consistency condition should be satisfied
d(I) = na , (4.462)
for all I ∈ Ωa , a ∈ ∆.
A simple example: for D = 11 supergravity with na = 4 the p-brane intersection d(I1 ∩ I2 ) = 3
is “forbidden” due to Restriction 3.
The solution mentioned above may be obtained by a straightforward substitution of the fields
(??), (??), (4.458) into equations of motion (2.63), (7.1308)-(7.1410) while formulas from the Ap-
pendix 2 are keeping in mind.
Electro-magnetic representation. Due to relation
¯
∗τ (I) = ε(I)δ(I)τ (I), (4.463)
CBPF-MO-003/02 73
where ∗ = ∗[g] is the Hodge operator on (M, g), I¯ = {0, . . . , n} \ I is “dual” set and δ(I) = ±1 is
¯ = δ(I)τ ({0, . . . , n}), the “electric brane” living on MI (see (7.1423))
defined by relation τ (I) ∧ τ (I)
may be interpreted also as a “magnetic brane” living on MI¯. The relation (??) may be rewritten
in the ”electromagnetic” form
X X
Fa = QaIe τ (I) + QaJm ∗ τ (J), (4.464)
I∈Ωae J∈Ωam
where Ωa = Ωae ∪ Ω∗am , Ωae ∩ Ω∗am = ∅, Ω∗ ≡ {J|J = I, ¯ I ∈ Ω}, and QaIe = QaI for I ∈ Ωae and
QaJm = QaJ¯ for J ∈ Ωam .
Here we consider some examples of the obtained solutions when ε0 = −1 and all εi = 1,
i = 1, . . . , n, i.e. “our space” (M0 , g 0 ) is pseudo-Euclidean space and the “internal spaces” (Mi , g i )
are Euclidean ones. We also put θa = 1 and na < D − 1 for all a ∈ ∆.
Solution with one p-brane. Let Ωa = {I}, λa = 0 for some a ∈ ∆ and Ωb are empty for all
b 6= a, b ∈ ∆. Equations (??) are satisfied identically in this case and (4.460) read
2Λ na − 1
ξi = + ε(I)Q2 [δIi − ], (4.465)
D−2 D−2
i = 0, . . . , n, where Q = QaI .
p-brane does not “live” in M0 . For I = {1, . . . , k}, 1 ≤ k ≤ n, we get ε(I) = 1 and
2Λ na − 1
ξ0 = ξk+1 = . . . = ξn = − Q2 , (4.466)
D−2 D−2
2Λ na − 1
ξ1 = . . . = ξk = + Q2 [1 − ]. (4.467)
D−2 D−2
For Λ = 0, Q 6= 0 we get ξ0 = ξk+1 = . . . = ξn < 0 and ξ1 = . . . = ξk > 0. These solutions
contain the solutions with the manifold
2Λ
ξ0 = − Q2 e2λa (ϕ) , (4.473)
D−2
2Λ
ξ1 = + Q2 e2λa (ϕ) . (4.474)
D−2
For Λ = 0 and Q 6= 0 we get the solution defined on the manifold M = AdSd × S d . For odd d the
form (??) is self-dual (see subsection 3.1). The solution describes a composite p-brane configuration
containing AdS5 × S 5 solution in IIB supergravity as a special case.
Near-horizon limit for A2 -dyon in D = 11 supergravity. Here we consider the dyon
solution (??), (3.128) in D = 11-supergravity.
Let g 0 = dR ⊗ dR + R2 ĝ[S 2 ], H = C + M
R
, where C and M are constants and g[S 2 ] is the metric
on S 2 . For C = 1 the 4-dimensional section of the metric describes extremely charged Reissner-
Nordström black hole of mass M in the region out of the horizon: R > 0. Now we put C = 0 and
M = 1 (i.e. the so-called “near-horizon” limit is considered). We get the solution
γ0 − γ = f (X), N = ef , (4.478)
where f : M → R is a smooth function. Then we obtain the Lagrange system with the Lagrangian
µ f
Lf = e GÂB̂ (X)Ẋ Â Ẋ B̂ − µ e−f Vw (4.479)
2
and the energy constraint
µ f
Ef = e GÂB̂ (X)Ẋ Â Ẋ B̂ + µ e−f Vw = 0. (4.480)
2
CBPF-MO-003/02 75
Using the standard prescriptions of (covariant and conformally covariant) quantization (see, for
example, [55, 66, 67]) we are led to the Wheeler-DeWitt (WDW) equation
µ ¶
1 £ f ¤ a £ f ¤
Ĥ Ψ ≡ − ∆ e G + R e G + e µVw Ψf = 0,
f f −f
(4.481)
2µ µ
where
N −2
a = aN = , N = n + l + |S|. (4.482)
8(N − 1)
Here Ψf = Ψf (X) is the so-called “wave function of the universe” corresponding to the f -gauge
(??) and satisfying the relation
(∆[G1 ] and R[G1 ] denote the Laplace-Beltrami operator and the scalar curvature corresponding to
G1 , respectively).
For the scalar curvature of minisupermetric (??) we get [179]
X X 0
R[G] = − (U s , U s ) − (U s , U s ). (4.484)
s∈S s,s0 ∈S
Here Ĥ ≡ Ĥ f =0 and Ψ ≡ Ψf =0 .
4.4.2 Quantum solutions with one curved factor space and orthogonal U s
Here as in Subsect. 4.2.2 we put Λ = 0, ξ1 6= 0, ξ2 = . . . = ξn = 0, and 1 ∈ / Is , s ∈ S, i.e. the
space M1 is curved and others are Ricci-flat and all “brane” submanifolds do not contain M1 . We
0
also put orthogonality restriction on the vectors U s : (U s , U s ) = 0 for s 6= s0 and Ks = (U s , U s ) 6= 0
for all s ∈ S.
CBPF-MO-003/02 76
zA = S AB σB , (z A ) = (z 1 , z a , z s ) (4.490)
as follows X
Ĝ = ĜAB dσ A ⊗ dσ B = −dz 1 ⊗ dz 1 + ηs dz s ⊗ dz s + dz a ⊗ dz b ηab , (4.491)
s∈S
s = (as , vs , Is ) ∈ S.
From (??), (??)-(4.493) we get
µ ¶2 X µ ¶
∂ ∂ ∂
ab qs z s ∂ −qs z s ∂
∆[G] = − +η + ηs e e (4.494)
∂z 1 ∂z a ∂z b s∈S ∂z s ∂z s
X µ ¶2
2qs z s ∂
+ εs e . (4.495)
s∈S
∂Φs
We are seeking the solution to WDW equation (??) by the method of the separation of variables,
i.e. we put à !
Y s a
Ψ∗ (z) = Ψ1 (z 1 ) Ψs (z s ) eIm Ps Φ eIm pa z . (4.497)
s∈S
s ∈ S, and X
2E1 + η ab pa pb + 2 Es + 2aR[G] = 0, (4.500)
s∈S
Using the relations from Appendix 5 we obtain linearly independent solutions to (??) and (4.499)
respectively
à !
p e q1 z 1
Ψ1 (z 1 ) = Bω1 1 −wµ2 ξ1 d1 , (4.501)
q1
µ ¶
s qs z s /2 s
p e qs z s
Ψs (z ) = e Bωs ηs εs Ps2 , (4.502)
qs
where r
p 1
ω1 = 2E1 /q1 , ωs = − 2ηs Es νs2 , (4.503)
4
s ∈ S and Bω1 , Bωs = Iω , Kω are the modified Bessel function.
The general solution of the WDW equation (??) is a superposition of the ”separated” solutions
(??):
XZ
Ψ(z) = dpdP dEC(p, P, E, B)Ψ∗ (z|p, P, E, B), (4.504)
B
where potential VQ is defined in (??). This equation describes quantum cosmology with classical
fields of forms and quantum scale factors and dilatonic fields. Such approach is equivalent to the
scheme of quantization of multidimensional perfect fluid considered in [72].
Eq. (??) is readily solved in the orthogonal case (see Appendix 5). The solutions are given by
the following modifications in eqs. (??), (4.502) and (??), respectively,
X
2E1 + η ab pa pb + 2 Es = 0, (4.506)
s∈S
µ s¶
s
p eqs z
Ψs (z ) = Bωs s 2
η s ε s Qs , (4.507)
qs
p
ωs = −2ηs Es νs2 . (4.508)
This solution for the special case with one internal space (n = 1) and non-composite p-branes
was considered in [159].
CBPF-MO-003/02 78
w = 1, M1 = S d1 , g 1 = dΩ2d1 , (5.509)
where dΩ2d1 is the canonical metric on a unit sphere S d1 , d1 ≥ 2. In this case ξ 1 = d1 − 1. We put
equivalent to Z u
t − t0 = ± dūΦ(ū), (5.515)
u0
Here we restrict ourselves to solutions with C1 > 0 and linear asymptotics at infinity
q s = −β s u + β̄ s + o(1), (5.517)
0 0 0
for all s ∈ S. Remind that Ass0 = 2(U s , U s )/(U s , U s ) is a quasi-Cartan matrix. In this case the
energy (3.233) reads
1 X 0
ET L = hs Ass0 β s β s . (5.519)
4 s,s0 ∈S
CBPF-MO-003/02 79
Remark 9. For positive definite matrices (hs Ass0 ) and (hαβ ) we get from (??) and (??): ET L ≥
0, C1 ≥ 0. (For the extremal case ET L = C1 = 0 see Subsect. 5.5. below.) According to Lemma 2
from [228] black hole solutions can only exist for C1 ≥ 0 and the horizon is then at u = ∞.
For the function (??) we get
where Φ0 6= 0 is constant, p X
β = c1 − c2 + C1 h1 + βs hs δ2Is , (5.521)
s∈S
1 1 −1
and h1 = (U , U ) = d1 /(1 − d1 ).
Horizon at u = +∞ takes place if and only if
β ≥ 0. (5.522)
UAr bA = 0, r = s, 1; s ∈ S, (5.524)
1 X 0
hs Ass0 bs bs + ĜAB bA bB = |h1 |, (5.525)
2 s,s0 ∈S
X 0
Ass0 bs > 0, (5.526)
s0 ∈S
X
f (b) ≡ b1 − b2 + bs hs δ2Is ≥ |h1 |. (5.527)
s∈S
Proposition 3. Let matrix (hαβ ) be positive definite. Then the point b = (bs , bA ) satisfying
relations (??)-(5.527) exists only if
2 ∈ Is , ∀s ∈ S, (5.528)
(i. e. all p-branes have a common time direction t) and is unique: b = b0 , where
X
bA
0 = −δ2
A
+ h 1 U 1A
+ hs bs0 U sA , (5.529)
s∈S
X 0
bs0 = 2 Ass , (5.530)
s0 ∈S
0 0 0 0
where s ∈ S, A = (i, α), and the matrix (Ass ) is inverse to the matrix (Ass0 ) = (2(U s , U s )/(U s , U s )).
signature. Due to (U 1 , U 1 ) < 0, the intersection of the hyperboloid (5.525) with the hyperplane
UA1 z A = 0 is ellipsoid. Its intersection with the ”planes” UAs z A = 0, s ∈ S, give us an ellipsoid,
coinciding with E.
Let us consider a function f| : E → R that is a restriction of the linear function (5.527) on E.
Let b∗ ∈ E be a point of maximum of f| . Using the conditional extremum method and the fact that
E is an ellipsoid we prove that
X
bA
∗ = −δ A
2 + h 1 U 1A
+ hs bs∗ U sA , (5.531)
s∈S
X 0
bs∗ = 2 Ass δ2Is0 , (5.532)
s0 ∈S
s ∈ S. Then, the points b∗ and −b∗ are the points of maximum and minimum, respectively, for the
function f| defined on the ellipsoid E. Since f (b∗ ) = |h1 |, the only point satisfying the restriction
f (b) ≥ |h1 | is b = b∗ . From (5.526) we get
X 0
Ass0 bs = 2δ2Is > 0 ⇐⇒ 2 ∈ Is , (5.535)
s0 ∈S
q s (0) = 0, (5.537)
A = (i, α), s ∈ S. These relations guarantee the asymptotical flatness (for R → +∞) of the
(2 + d1 )-dimensional section of the metric.
Let us denote
s
Hs = fs e−µ̄b0 u , (5.538)
s ∈ S. Then, solutions (??)-(??) may be written as follows
³Y ´n
g= Hs2hs d(Is )/(D−2) F −1 dR ⊗ dR + R2 dΩ2d1 (5.539)
s∈S
³Y ´ n ³Y
X ´ o
− Hs−2hs F dt ⊗ dt + Hs−2hs δiIs ĝ i , (5.540)
s∈S i=3 s∈S
Y hs χs λα
exp(ϕα ) = Hs as
, (5.541)
s∈S
X
a
F = δaas F s , (5.542)
s∈S
CBPF-MO-003/02 81
where à !
Qs Y −Ass0
F s = − d1 Hs0 dR ∧ τ (Is ), (5.543)
R s0 ∈S
s ∈ Se ,
F s = Qs τ (I¯s ), (5.544)
s ∈ Sm . Here Qs 6= 0, hs = Ks−1 ; s s
(Ks = (U , U ) 6= 0), s ∈ S, and the quasi-Cartan matrix (Ass0 ) is
non-degenerate.
Functions Hs > 0 obey the equations
µ ¶ Y −A 0
d1 d d1 F dHs
R R = Bs Hs0 ss , (5.545)
dR Hs dR 0
s ∈S
s ∈ S, where Bs = εs Ks Q2s 6= 0. These equations follow from Toda-type equations (3.223) and the
definitions (??) and (??).
It follows from (5.517), (??), (??) and (??) that there exist finite limits
Hs → Hs0 6= 0, (5.546)
d¯ d¯
for R → 2µ, s ∈ S. In this case the metric (??) has a regular horizon at R = 2µ. From (??) we
get
Hs (R = +∞) = 1, (5.547)
s ∈ S.
The Hawking ”temperature” corresponding to the solution is (see also [154, 155] for orthogonal
case) found to be
d¯ Y
−hs
TH = 1/d¯ Hs0 , (5.548)
4π(2µ) s∈S
where Hs0 are defined in (??)
The boundary conditions (??) and (5.547) play a crucial role here, since they single out, generally
speaking, only few solutions to eqs. (??). Moreover, for some values of parameters µ = µ̄/d, ¯ εs
2
and Qs the solutions to eqs. (??)-(5.547) do not exist. Indeed, from (3.224), (3.233), (??), (??),
(5.530), (??) and (??) we get
X 1 X X1
hs Ass0 q˙s (0)q˙s0 (0) +
0
ET L = µ̄2 hs Ass = εs Q2s . (5.549)
s,s0 ∈S
4 s,s0 ∈S s∈S
2
Let P
the matrix (Ass0 )Pbe positive-definite and all hs > 0). Then ET L > 0 and
0
2µ̄2 s,s0 ∈S hs Ass ≥ s∈S εs Q2s . If the parameters obey the relation
P 0 P
0 < 2µ̄2 s,s0 ∈S hs Ass < 2 2
s∈S εs Qs , e.g. for εs = +1 and big enough Qs , the solution under
consideration does not exist.
We note that the solution to eqs. (??)-(5.547) may not be unique. The simplest example occurs
in the case of one p-brane, when hs > 0, εs = +1 and µ̄2 hs > Q2s . In this case we have two solutions
to (??)-(5.547) corresponding to two possible values of q˙s (0).
Thus, we obtained a family of black hole solutions up to solutions of radial equations (??) with
the boundary conditions (??) and (5.547). In the next sections we consider several exact solutions
to eqs. (??)-(5.547).
Remark 10. Let Mi = R and g i = −dt̄ ⊗ dt̄ for some i ≥ 3. Then the metric (??) has no
¯
a horizon with respect to the “second time” t̄ for Rd → 2µ. Thus, we are led to a “single-time”
theorem from [228]. Relation (??) from Proposition 3 coincides with the “no-hair” theorem from
[228].
CBPF-MO-003/02 82
where Hs (z) > 0, F = 1 − 2µz, µ > 0, z = R−d ∈ (0, (2µ)(−1) and B̄s = Bs /d¯2 6= 0. Eqs. (??) and
¯
(5.547) read
s ∈ S.
It seems rather difficult to find the solutions to a set of eqs. (??)-(5.552) for arbitrary values
of parameters µ, B̄s , s ∈ S, and quasi-Cartan matrices A = (Ass0 ). But we may expect a drastical
simplification of the problem under consideration for certain class of parameters and/or A-matrices.
In general we may try to seek solutions of (??) in a class of functions analytical in a disc |z| < L
and continuous in semi-interval 0 < z ≤ (2µ)−1 . For |z| < L we get
∞
X
Hs (z) = 1 + Ps(k) z k , (5.553)
k=1
(k)
where Ps are constants, s ∈ S. Substitution of (??) into (??) gives us an infinite chain of relations
(k)
on parameters Ps and B̄s . In general case it seems to be impossible to solve this chain of equations.
Meanwhile there exist solutions to eqs. (??)-(5.552) of polynomial type. The simplest example
0
occurs in orthogonal case [136, 153, 154, 179, 155]: (U s , U s ) = 0, for s 6= s0 , s, s0 ∈ S. In this case
(Ass0 ) = diag(2, . . . , 2) is a Cartan matrix for semisimple Lie algebra A1 ⊕ . . . ⊕ A1 and
Hs (z) = 1 + Ps z, (5.554)
with Ps 6= 0, satisfying
Ps (Ps + 2µ) = −B̄s , (5.555)
s ∈ S.
In [167, 3, 223] this solution was generalized to a block-orthogonal case (3.86), (3.87). In this
case (??) is modified as follows
s
Hs (z) = (1 + Ps z)b0 , (5.556)
where bs0 are defined in (5.530) and parameters Ps and are coinciding inside blocks, i.e. Ps = Ps0
for s, s0 ∈ Si , i = 1, . . . , k. Parameters Ps 6= 0 satisfy the relations (??) and parameters B̄s are also
coinciding inside blocks, i.e. B̄s = B̄s0 for s, s0 ∈ Si , i = 1, . . . , k. In this case Hs are analytical in
the disc |z| < L, where L = min(|Ps |−1 , s ∈ S).
Let (Ass0 ) be a Cartan matrix for a finite-dimensional semisimple Lie algebra G. In this case all
powers in (5.530) are natural numbers coinciding with the components of twice the dual Weyl vector
in the basis of simple coroots [187] (see Appendix 3) and hence, all functions Hs are polynomials,
s ∈ S.
CBPF-MO-003/02 83
Conjecture 1. Let (Ass0 ) be a Cartan matrix for a semisimple finite-dimensional Lie algebra
G. Then the solution to eqs. (??)-(5.552) (if exists) is a polynomial
ns
X
Hs (z) = 1 + Ps(k) z k , (5.557)
k=1
(k) P 0 (n )
where Ps are constants, k = 1, . . . , ns ; ns = 2 s0 ∈S Ass ∈ N and Ps s 6= 0, s ∈ S.
In extremal case (µ = +0) an a analogue of this conjecture was suggested previously in [159].
... (5.561)
m
β = µ̄m = w2 + . . . + wm+1 . (5.562)
These relations imply
wm+1 = µ̄m, wm = µ̄(m − 2), . . . , w1 = −µ̄m, (5.563)
or, wj = (2j − m − 2)µ̄, j = 1, . . . , m + 1. It follows from (??) and (??) that
s
fs = e−q = αs(0) ens µ̄u + αs(1) e(ns −2)µ̄u + . . . + αs(ns ) e−ns µ̄u , (5.564)
(k) (ns )
where αs are constants, k = 1, . . . , ns , αs 6= 0. Hence, due to (??) and (??) we obtain the
relations
s −n
Hs = e−q s µ̄u
= αs(0) + αs(1) F + . . . + αs(ns ) F ns , (5.565)
(n ) (n )
equivalent to (??) with αs s = Ps s 6= 0, s = 1, . . . , m. Thus, the Conjecture 1 is proved for the
Lie algebras G = Am , m ≥ 1.
Now we prove the Conjecture 1 for simple Lie algebras Cm+1 = sp(m + 1), m ≥ 1. (Remind
that for m = 1: C2 = B2 = so(5)). The Cartan matrix for the Lie algebra Cm+1 (m ≥ 1) reads
2 −2 0 . . . 0 0
−1 2 −1 . . . 0 0
0 −1 2 . . . 0 0
(Ass0 ) = (5.566)
. . . . . . . . . . . . . . . . . . . . . . . . . . .
0 0 0 . . . 2 −1
0 0 0 . . . −1 2
CBPF-MO-003/02 84
s, s0 = 0, . . . , m. The set of equations (??) with the Cartan matrix (??) and s = 0, . . . , m, may be
embedded into a set of equations (??) corresponding to the Cartan matrix of the Lie algebra A2m+1
(see (??)) with s = −m, . . . , 0, . . . , m, if the following identifications : B̄−k = B̄k and H−k = Hk ,
k = 1, . . . , m, are adopted. It may be verified using Appendix 3 that ns (Cm+1 ) = ns (A2m+1 ),
s = 0, . . . , m. This proves the Conjecture 1 for Cm+1 , since it was proved for A2m+1 .
Thus, in the A2 -case the solution is described by relations (??)-(5.544) with S = {s1 , s2 } (iden-
tified with {1, 2}), intersection rules
where symbol ∆(s, s0 ) is defined in (3.108) and K = Ksi = (U si , U si ) 6= 0; functions Hsi = Hi are
¯
defined by relations (??) and (??)-(??) with z = R−d , i = 1, 2.
√
dA = 2F .
For Q2 → 0 we get the black hole version of Dobiash-Maison solution from [231] and for Q1 → 0
we are led to the black hole version of Gross-Perry-Sorkin monopole solution from [232, 233], see
[235]. The solution coincides with Gibbons-Wiltshire dyon solution [234]. Our notations are related
√
to those √from ref. [234], as following : H1 R2 = B, H2 R2 = A, R2 − 2µR = ∆, Q1 = √ 2q,
2 2 2 2 2
Q2 = − 2p, R − µ = r − m, µ = m + d − p − q , (P2 − P1 )/2(P2 + 1) = d/(d − 3m)).
(For general spherically symmetric configurations see also ref. [23].) In [236] the KK dyon solution
[234] was used for constructing the dyon solution in D = 11 supergravity for flat g 3 and g 4 and its
rotating version.
Let us imagine that some real astrophysical objects (e.g. stars) may be described (or approxi-
mated) by the 4-dimensional physical metric (??), i.e. they are traces of extended multidimensional
objects (charged p-brane black holes).
For post-Newtonian approximation we restrict ourselves by the first two powers of 1/R, i. e.
(2)
Ps Ps 1
Hs = 1 + + 2 + o( 3 ), (5.588)
R R R
for R → +∞, s ∈ S.
Introducing a new radial variable ρ by the relation R = ρ(1 + (µ/2ρ))2 , (ρ > µ/2), we may
rewrite the metric (??) in the 3-dimensional conformally-flat form
½ µ ¶4 ¾
(4) (1 − (µ/2ρ))2 µ i j
g = U −U1 dt ⊗ dt + 1 + δij dx ⊗ dx , (5.589)
(1 + (µ/2ρ))2 2ρ
and
X µ ¶
1 2 (2) d(Is )
β−1= hs (Ps + 2µPs − 2Ps ) 1 − , (5.593)
2(GM )2 s∈S D−2
µ ¶
1 X d(Is )
γ−1=− hs Ps 1 − 2 . (5.594)
GM s∈S D−2
Now, we show that in general case, like in [223], the parameter β is defined by the charges squared
Q2s of p-branes (or, more correctly, by the charge densities squared) and signature parameters εs .
Indeed, from (??) we get in zeroth order of 1/R-decomposition:
β > 1. (5.597)
³Y ´ n ³Y
X ´ o
− Hs−2hs dt ⊗ dt + Hs−2hs δiIs ĝ i , (5.599)
s∈S i=3 s∈S
and the relations for scalar fields and fields of forms (5.541)-(5.544) are unchanged. Here Hs =
¯
Hs (z) > 0, z = R−d ∈ (0, +∞) and the following relations are satisfied
d2 Y −A 0
2
ln H s = B̄ s Hs0 ss , Hs (+0) = 1, (5.600)
dz s0 ∈S
d¯2 X d d X Y −A 0
ET L = hs Ass0 ( ln Hs ) ln Hs0 + As Hs0 ss = 0, (5.601)
4 s,s0 ∈S dz dz s∈S s0 ∈S
CBPF-MO-003/02 88
where B̄s = Bs /d¯2 6= 0, and Bs , As are defined in (3.224), s ∈ S. These solution may be obtained
as a special case of solutions from Subsect. 4.2.2 with
C1 = ET L = cA = 0, (5.602)
(k) (n )
where Ps are constants, k = 1, . . . , ns ; ns = bs0 are defined in (5.530) and Ps s 6= 0, s ∈ S.
(n )
This conjecture was suggested (in fact) previously in [159]. The coefficients Ps s = Cs > 0 may
be calculated by substitution of asymptotical relations
s
Hs (z) ∼ Cs z b0 , z → +∞, (5.604)
s ∈ S. We note that the asymptotical relations (??) are satisfied in a more general case, when
B̄s bs0 < 0, s ∈ S.
Let us consider the metric (??) with Hs obeying asymptotical relations (??). We have a horizon
for R → +0, if
X 1
ξ= hs bs0 − ≥ 0, (5.606)
s∈S
d0 − 2
where d0 = d1 + 1. This relation follows from the requirement of infinite time propagation of light
to R → +0.
For flat internal spaces Mi = Rdi , i = 3, . . . , n, we get for the Riemann tensor squared
(Kretschmann scalar) ¿from Appendix 1
nd C ≥ 0 (C = const). Due to (??) the metric (??) with flat internal spaces has no curvature
singularity when R → +0, if
η ≥ 0. (5.609)
For hs bs0 > 0, d(Is ) < D − 2, s ∈ S, we get η < ξ and relation (??) single out extremal charged
black holes with p-branes and flat internal spaces.
CBPF-MO-003/02 89
where
Y −Ass0
F s = Qs ( Hs0 )dH ∧ τ (Is ), s ∈ Se , (5.614)
s0 ∈S
where X is finite non-empty subset X ⊂ M0 and all qb > 0 for b ∈ X. For flat internal spaces
Mi = Rdi , i = 3, . . . , n, and non-negative indices η and ξ (see (??) and (??)) the solution describes
a set of |X| extremal p-brane black holes. Here relations H(x) → 0 for |x| → +∞ and Hs (+0) = 1
(s ∈ S) imply the asymptotical flatness of the (1 + d0 )-dimensional section of the metric. A black
hole corresponding to a “point” (horizon) b ∈ X carries brane charges Qs qb , s ∈ S. Since the
solution is invariant under the replacement of parameters:PQs 7→ αQs , qb 7→ Qs /α, α > 0, b ∈ X,
s ∈ S, we may normalize parameters qb by the restriction b∈X qb = 1.
solutions in arbitrary dimensions - see e.g. [3,155,214,163,223,239,334] and references therein. They
are important in connection with studies of processes at early stages of the Universe, counts of
micro-states in BH thermodynamics and now especially due to new developments in M-theory [335]
related to the AdS/CFT correspondence [205]. For recent reviews of this rapidly developing field
see, e.g., [336, 337].
BH stability studies have a long history, of which we will only mention (more or less arbi-
trarily) some milestones, concerning spherically symmetricbackgrounds. Regge and Wheeler [338]
considered the stability of the Schwarzschild space-time and developed the formalism of spheri-
cal harmonics for metric perturbations. Vishveshwara [339] finally proved the linear stability of
Schwarzschild BHs; Moncrief [340] did the same for Reissner-Nordström ones. BHs with a con-
formally coupled scalar field were shown to be unstable under spherically symmetricperturbations
[341], as well as minimally coupled scalar field configurations in general relativity possessing naked
singularities [342]. The monopole degree of freedom is present there due to the scalar field; it was
argued that monopole perturbations were most likely to be unstable due to the absence of cen-
trifugal terms in the effective potentials; catastrophic instabilities were indeed found and it was
unnecessary to study other multipoles. On the other hand, coloured BHs, containing non-Abelian
gauge fields, were shown to be, in general, unstable due to their sphaleronic degrees of freedom —
see [343] and references therein. A recent overview of 4-dimensional perturbation studies may be
found in Ref. [344].
For BHs in multidimensional theories of gravity the situation is more complex since, on the
one hand, there emerge new effective scalar fields (extra-dimension scale factors, sometimes called
moduli fields) in the external space-time, and, on the other, instabilities may be caused by waves
in extra dimensions. Instabilities of the latter kind were indeed found by Gregory and Laflamme
[345, 346] for a limited class of neutral and charged black strings and branes, having a constant
internal space scale factor. Furthermore, it was argued that compactification on a sufficiently small
length scale should prevent the onset of instability, and, moreover, that extremal black branes are
stable [347]. It was concluded that only very light BHs, whose horizon radii have the same order of
magnitude as their extra dimensions, manifest this form of instability.
It is therefore of interest to inquire whether or not there are other forms of instability, maybe
“more dangerous”, on more general backgrounds, containing nontrivial internal space structures
and/or several dilatonic scalars and brane charges. As was previously the case with backgrounds
containing effective scalar fields, it is natural to consider first the simplest, monopole perturbations.
Earlier we analyzed the stability of static, spherically symmetricsolutions to the Einstein-Maxwell-
scalar equations with a dilatonic type coupling between scalar and electromagnetic fields in D-
dimensional gravity [348,100].It was proved there that only BH configurations were stable under
linear spherically symmetricperturbations, while non-BH solutions turned out to be catastrophically
unstable. A similar result was obtained for dilatonic BHs with the inclusion of the Gauss-Bonnet
curvature term due to one-loop quantum corrections [349]. We will now show that in the simplest
case of a single charged black brane the solution is stable under linear spherically symmetricper-
turbations, whereas single-brane solutions with naked singularities are unstable. So the results of
[348,100] are generalized.
We also present a tentative consideration of multi-brane BHs and conclude that in cases when the
perturbation equations decouple, the stability conclusion is also valid. Two classes of such systems
are indicated, both characterized by certain relations among brane charges, such that, in terms
of Sect 3, the constituent vectors Y~ s form a single block of a block-orthogonal system (BOS) —
single-block BHs for short. Namely, the stability is proved for arbitrary two-brane single-block BHs
and multi-brane single-block BHs with mutually orthogonal vectors Y~ s (see the details in 5.6.6). For
CBPF-MO-003/02 91
many single-block configurations which do not belong to these classes, the stability can be proved as
well, but their properties require individual studies; see an example in 5.6.8, Eqs. (5.713)–(5.208).
There are, however, numerous multi-brane BHs for which decoupling is impossible and one may
expect that some of them show a new type of instability connected with mode interaction; a study
of these systems is in progress.
So, 5.6.2 describes the general features of the field model to be considered. 5.6.3 presents some
known static solutions, including BHs, on the basis of the target space V connected with dimensional
reduction. In 5.6.4 a truncated target space V, more appropriate for treating the perturbations, is
introduced, and wave equations for perturbations are derived. In 5.6.5 the stability properties of
single-brane configurations are deduced, while in 5.6.6 the stability of some multi-brane BHs under
spherically symmetricperturbations is established. 5.6.8 gives some examples from 11-dimensional
supergravity.
The word “stable” throughout here means “stable under linear spherically symmetricperturba-
tions”.
Here u is a radial coordinate ranging in Ru ⊆ R; ds20 = dΩ2 is the metric on a unit d0 -dimensional
sphere M 0 = S d0 ; t ∈ M 1 ≡ Rt is time; the metrics g i = ds2i of the “extra” factor spaces (i ≥ 2)
are assumed to be Ricci-flat and can have arbitrary signatures εi = sign g i ; |g| = | det gM N | and
M ...M
similarly for subspaces; Fs2 = Fs, M1 ...Mns Fs 1 ns ; λsa are coupling constants; ηs = ±1 (to be
i
specified later); s ∈ S, a ∈ A, where S and A are some finite sets. The “scale factors” eβ and the
scalars ϕa are assumed to depend on u and t only.
The F -forms should be also compatible with spherical symmetry. A given F -form may have
several essentially (non-permutatively) different components; such a situation is sometimes called
“composite p-branes”1 . For convenience, we will nevertheless treat essentially different components
of the same F -form as individual (“elementary”) F -forms. A reformulation to the composite ansatz,
if needed, is straightforward.
Each ns -form F = dA ≡ ∂[M1 AM2 ...Mns ] dz M1 . . . dz Mns is then associated with a certain subset I =
{i1 , . . . , ik } (i1 < . . . < ik ) of the set of numbers labelling the factor spaces: {i} = I0 = {0, . . . , n}.
1
There is an exception: two components, having only one noncoinciding index, cannot coexist since in this case
N
there emerge nonzero off-block-diagonal components of the energy-momentum tensor (EMT) TM , while the Einstein
tensor in the l.h.s. of the Einstein equations is block-diagonal. See more details in [147].
CBPF-MO-003/02 92
The forms Fs are naturally classified as electric (F eI ) and magnetic (FmI ) ones. By definition, the
potential AI of an electric form F eI carries the coordinate indices of the subspaces M i , i ∈ I and is
u-dependent (since only a radial component of the field may be nonzero). A magnetic form FmI is
built as a form dual to a possible electric one, and its nonzero components carry coordinate indices
def
of the subspaces M i , i ∈ I = I0 \ I, One can write:
We will make some more assumptions to assure that all F -forms behave like genuine electric or
magnetic fields in the physical subspace M phys = Ru × Rt × M 0 , namely:
By (i), the so-called quasiscalar forms [155,214] (forms with 1 6∈ Is , behaving as effective scalar
or pseudoscalar fields in M phys ) are excluded. The reason for adopting (i) is that our interest here
is mostly in BHs which do not admit nonzero quasiscalar forms (the no-hair theorem for brane
systems [228].
Assumption (iii) holds if all extra dimensions are spacelike (εi = 1, i ≥ 2) and in (7.1305) all
ηs = 1. In more general models, with arbitrary εi , (iii) holds if
def
Y
η eIs = −ε(Is ), ηmIs = ε(I s ), ε(I) = εi . (5.625)
i∈I
We will consider static configurations and their small (linear) time-dependent perturbations. It
turns out, however, that under the above assumptions the Maxwell-like field equations for Fs may
be integrated in a general form for their arbitrary dependence on u and t. Indeed, for an electric
m-form Fs (s = eI, m = d(Is ) + 1) the field equations due to (5.110)
µ ¶³ ´
∂u p a
FsutM3 ...Mm |g| e2λsa ϕ = 0 (5.626)
∂t
Evidently, the expressions (5.119) and (5.120) differ only in the signs before λsa and the signature-
dependent prefactors ε. Due to (5.117), their energy-momentum tensors (EMTs) coincide up to the
replacement λsa → −λsa , and their further treatment is quite identical. In what follows we therefore
mostly speak of electric forms, but the results are easily reformulated for any sets of electric and
magnetic forms. We also assume that all Qs 6= 0.
where k is an integration constant. With (5.123) the D-dimensional line element may be written in
def
the form (d = d0 − 1)
½ ¾ Xn
e−2σ1 /d du2 2β 1 2 2β i
ds2D = 2
+ dΩ 2
− e dt + e ds2i . (5.632)
[ds(k, u)] 2/d [ds(k, u)] i=2
The u coordinate is defined for 0 < u < umax where u = 0 corresponds to spatial infinity while umax
may be finite or infinite depending on the form of a particular solution.
The remaining set of unknowns β i (u), ϕa (u) (i = 1, . . . , n, a ∈ A) can be treated as a real-
valued vector function xA (u) (so that {A} = {1, . . . , n} ∪ A) in an (n + |A|)-dimensional vector
space V (target space). The field equations for xA can be derived from the Toda-like Lagrangian
X n
X 2 X
σ1, u
L= GAB xA B
u xu + Q2s e2ys (u) ≡ di (βui )2 + + δab ϕau ϕbu + Q2s e2ys (u) (5.633)
s i=1
d0 − 1 s
where δiI = 1 if i ∈ I and δiI = 0 otherwise. The contravariant components and scalar products of
the vectors Y~ s are found using the matrix GAB inverse to GAB :
µ ij ¶ ³ ´
AB δ /di − 1/(D − 2) 0 A d(Is )
(G ) = , (Y s ) = δiIs − , −λ sa ; (5.637)
0 δ ab D−2
d(Is )d(Is0 )
Ys,A Ys0 A ≡ Y~ s Y~ s0 = d(Is ∩ Is0 ) − + λas λas0 . (5.638)
D−2
where the vectors of integration constants ~c and ~c are orthogonal to all Y~ s : cA Ys,A = cA Ys,A = 0,
or
ci di δiIs − ca λsa = 0, ci di δiIs − ca λsa = 0. (5.644)
Namely, assuming that some of the functions ys (u) (5.128) coincide, one obtains the so-called BOS
solutions [214], where the number of independent charges coincides with the number of different
functions ys (u).
Indeed, suppose [214] that the set S splits into several non-intersecting non-empty subsets,
[
S= Sω , |Sω | = m(ω), (5.645)
ω
such that the vectors Y~ µ(ω) (µ(ω) ∈ Sω ) form mutually orthogonal subspaces Vω ⊆ V:
Then one has the following solution to the equations of motion (5.131), (5.126):
X Yω A
xA = yω (u) + cA u + cA , (5.648)
ω
Yω2
def
X
e−2yω = qbω Yω2 s2 (hω , u + uω ), qbω = Q2µ , (5.649)
ω
X h2 sign hω d0
ω
E = 2
+ ~c 2 = k 2 sign k (5.650)
ω
Y ω d0 − 1
where hω , uω , cA and cA are integration constants; cA and cA are constrained by the orthogonality
relations (5.136) (so that the vectors ~c and ~c are orthogonal to each individual vector Y~ s ∈ V).
Eqs. (5.139) form a set of linear algebraic equations with respect to the “charge factors” aν =
P
q ω , satisfying the condition ω aµ = 1. A solution to (5.139) for given Y~ µ can contain some
Q2ν /b
aµ < 0; according to [214], this would mean that such a p-brane is “quasiscalar”, violating the
assumption (5.622). Solutions with such branes are possible but are rejected here since they do not
lead to black holes. Furthermore, if a solution to (5.139) gives aµ = 0 for some µ ∈ Sω , this means
that the block cannot contain such a p-brane, and then the consideration may be repeated without
it.2
The function yω (u) is equal to yµ(ω) (u) = Yµ(ω),A xA , which is, due to (5.139), the same for all
µ ∈ Sω . The BOS solution generalizes the OS one, (5.133), (5.134): the latter is restored when each
block contains a single F -form.
Both kinds of solutions are asymptotically flat, and it is natural to normalize the functions ys (u)
and yω (u) by the condition ys (0) = 0 or yω (0) = 0, so that the constants us and uω are directly
related to the charges.
Other solutions to the equations of motion are known, connected with Toda chains and Lie
algebras [163, 334, 3, 239].
2 ~ ω solving Eqs. (5.139) is the altitude of the pyramid formed by the vectors Y
Geometrically, the vector Y ~ µ , µ ∈ Sω
with a common origin. The condition aµ > 0 means that this altitude is located inside the pyramid, while aµ = 0
means that the altitude belongs to one of its faces.
CBPF-MO-003/02 96
Black-hole solutions
Black holes (BHs) are distinguished among other spherically symmetricsolutions by the existence
of horizons instead of singularities in the physical d0 +2-dimension space M phys ; the extra dimensions
and scalar fields are also required to be well-behaved on the horizon to provide regularity of the
D-dimensional metric. Thus BHs are described by the above solutions under certain constraints
upon the integration constants. Namely,p for the BOS solution (5.139)–(5.142), requiring that all
i 1
the scale factors eβ (except eβ = |gtt | which should tend to zero) and scalars ϕa tend to finite
limits as u → umax , we get [155]:
X
hω = k > 0, ∀ ω; cA = k Yω−2 Yω A − kδ1A (5.651)
s
where A = 1 corresponds to i = 1 (time). The constraint (5.135) then holds automatically. The
value u = umax = ∞ corresponds to the horizon. The same condition for the OS solution (5.133)–
(5.136) is obtained by replacing ω 7→ s.
Under the asymptotic conditions ϕa → 0, β i → 0 as u → 0, after the transformation
2k def
e−2ku = 1 − , d = d0 − 1 (5.652)
drd
the metric (5.124) for BHs and the corresponding scalar fields may be written as
³Y ´h ³2k ´ Y −2/Yω2 ³ dr2 ´ Xn Y i
Aiω
ds2D = HωAω 2
−dt 1 − Hω + 2 2
+ r dΩ + 2
dsi Hω ;
ω
drd ω 1 − 2k/(drd ) i=2 ω
X 1 X OS
X λsa
ϕa = − 2
ln Hω aµ λµa = − ln Hs , (5.654)
ω
Y ω ω s
Ys2
OS
where = means “equal for OS, with ω 7→ s”, and Hω are harmonic functions in R+ × S d0 :
def p
Hω (r) = 1 + pω /(drd ), pω = k 2 + qbω Yω2 − k. (5.655)
The subfamily (5.143), (5.145)–(5.147) exhausts all BOS BH solutions; the OS ones are obtained
in the special case of each block Sω consisting of a single element s.
The above relations describe the so-called non-extremal BHs. Extremal ones, corresponding to
minimum black hole mass for given charges (the so-called BPS limit), are obtained in the limit k → 0.
The same solutions follow directly from (5.142)–(5.141) under the conditions hω = k = cA = 0. For
k = 0, the solution is defined in the whole range r > 0, while r = 0 in many cases corresponds
to a naked singularity rather than an event horizon, so that we no more deal with a black hole.
However, in many other important cases r = 0 is an event horizon of extremal Reissner-Nordström
type, with an AdS near-horizon geometry; some examples are mentioned in the Appendix.
Other families of solutions, mentioned at the end of the previouis section, also contain BH
subfamilies. The most general BH solutions are considered in [239].
CBPF-MO-003/02 97
As in Refs. [348, 100], it is helpful to pass to the Einstein frame in the physical (d0 + 2)-
dimensional space-time M phys = Ru × M 0 × M 1 . The action (5.110) is then rewritten in terms of
the metric gµν , the d0 + 2-dimensional part of gM N , and is transformed to the Einstein frame in
M phys with the metric
g µν = e2σ2 /d0 gµν . (5.657)
The electric ns -forms are re-parametrized as follows:
1 2 1 a
FutM3 ...Mns = F ut , Fs = F µν F µν e−2σ(Js ) = −Q2s e−2λsa ϕ , (5.658)
s ns ! 2 s s
where the indices M3 , . . . Mns belong to Js ; here and henceforth the indices µ, ν are raised and
lowered using the metric g µν ; in the last equality the solution (5.120) and the positive energy
assumption (5.624) are taken into account.
The action (5.110) is written in terms of g µν and F µν as follows (up to a constant prefactor,
s
connected with the volume of extra dimensions, and a subtracted total divergence):
Z p n X n
d0 +2 1
S = d z |g| R[g] − (∂σ2 )2 − di (∂β i )2
d0 i=2
M phys
1X a
o
− δab (∂ϕa , ∂ϕb ) − F µν F µν e2σ2 /d0 −2σ(Js )+2λsa ϕ
2 s∈S s s
Z n o
p 1X K
= dd0 +2 z |g| R[g] − HKL (∂xK , ∂xL ) − F µν F µν e−2Zs,K x (5.659)
2 s∈S s s
M phys
defines a positive-definite metric in the vector space V (truncated target space) parametrized by the
variables (xK ) = (β 2 , . . . , β n ; ϕa ); the constant vectors Z s ∈ V are characterized by the components3
³ di ´ ³ d(Is ) ´
(Zs,K ) = di δiJs − , −λsa , (Zs K ) = (H KL Zs,L ) = δiJs − , −λsa (5.661)
d0 D−2
3
We will use the indices K, L for quantities specified in V to distinguish them from those in V where the indices
A, B are used; vectors in V are marked with overbars, those in V by arrows. Scalar products are written as Y ~Z~ =
A B K L
GAB Y Z (as before) and Y Z = HKL Y Z .
CBPF-MO-003/02 98
The truncated target space V may be considered as the hyperplane x1 = −σ2 /d0 in V, with the
metric HKL induced by GAB (5.127). The components H KL turn out to be the same as GAB for
i 6= 1; the components Ys A and Zs K coincide in the same manner. It is easy to find that for vectors
whose Is satisfy (5.656) (which is always the case in the present paper),
d0 − 1
Y~ s Y~ s0 = Z s Z s0 + , (5.663)
d0
whence it follows that, first, when different Y~ s are mutually orthogonal in V, the corresponding Z s
are never mutually orthogonal in V; second, for any Y~ s whose Is 3 1 one has
d0 − 1
Y~ s Y~ s ≡ Ys2 > . (5.664)
d0
Wave equations
The action (5.659) may be used to obtain the equations governing small spherically symmet-
ricperturbations of static solutions. The metric (5.657) in M phys will be written in the form
where “E” stands for the Einstein frame and α(u, t), β(u, t), γ(u, t) are connected with the corre-
sponding quantities from (5.111) as follows:
Since the field equations for the F -forms have been integrated — see (5.119) and (5.658), the
remaining unknowns are α, β, γ and xK (that is, β i , i ≥ 2, and ϕa ). In what follows we will write
where δα is a small perturbation, and similarly for other unknowns. We accordingly preserve only
terms linear in δα and similar quantities and in time derivatives. The field equations may be written
in the form
X
[g] xK = Zs K Q2s e−2d0 β+2Z s x , (5.667)
s
1 ν λ def ν
Rνµ = Tµν − δ T = T̃ µ (5.668)
d0 µ λ
where [g] = g µν ∇µ ∇ν is the D’Alembert operator, while for the nonzero components of the EMT
Tµν corresponding to (5.659) one has (no summing in µ)
X ³ 1 1 1 1´
T̃ µµ = e −2α
HKL xK L
u xu diag(0, 1, 0, . . . , 0) − Q2s e−2α+2γ+2Z s x diag 1 − , − , − ..., − ,
s
d0 d0 d0 d0
T̃ ut = HKL xK L
u xt (5.669)
CBPF-MO-003/02 99
where xu = ∂u x and xt = ∂t x; the first and second places under the symbol “diag” belong to t and
u, respectively.
As in our previous papers on stability, we use the coordinate freedom in the perturbed space-time
and put
δβ ≡ 0 (5.670)
but preserve the harmonic u coordinate condition in the unperturbed (static) space-time4 . Then
Eqs. (5.667) and (5.668) give
X
L̂ δxK + xK
u (δγ u − δα u ) − 2x K
uu δα = 2 Q2s e2γ+2Z s x Zs K Zs,L δxL ; (5.671)
s
def 2d0 β
L̂ = − e ∂tt + ∂uu ;
d0 δαt = xu δxt ; (5.672)
d0 βu (δαu − δγu ) = 2xuu δx − 2βuu δα, (5.673)
where (5.671) follows from the (ut ) component of (5.668) and (5.673) from one of the angular
components of (5.668); we have also used the equations valid for static systems, in particular,
(5.131), where, according to the definitions of Y~ and Z, ys (u) = Y~ s ~x = γ + Z s x. Integrating (5.672)
in t and omitting the emerging arbitrary function of u (since we neglect static perturbations), we
obtain
d0 δα = xu δx. (5.674)
Substituting δα from (5.674) and δαu − δγu from (5.673) into (5.671), we finally arrive at the set of
wave equations for the dynamical degrees of freedom in our system, represented by δxK :
1 d ³ xK
u xL,u
´ X
L̂δxK = 2P K L δxL , P KL = + Q2s e2ys (u) ZsK Zs,L . (5.675)
d0 du βu s
The stability problem is now reduced to a boundary-value problem for δxK (u, t). Namely, if
there exists a nontrivial solution to Eqs. (5.675) satisfying some physically reasonable conditions at
the ends of the range of u, such that |δxK | (at least some of them) grow unboundedly with t, then
the static system is unstable. Otherwise it is stable in the linear approximation.
The condition at infinity, u = 0, is evident: the perturbations must vanish,
dxK → 0 as u → 0. (5.676)
It is less evident at u = umax since some of the background static solutions are singular there. As
in Refs. [342, 100] and others, dealing with minimally coupled or dilatonic scalar fields, we will use
the minimal requirement providing the validity of the perturbation scheme, namely
When the background is regular, this condition requires that the perturbation should be regular as
well.
4
This coordinate is harmonic for both metrics (5.111) and (5.665); in the latter the coordinate condition has the
form α = d0 β + γ.
CBPF-MO-003/02 100
Decoupling cases
Eqs. (5.675) in general do not decouple. Even in the simplest case when there is only one
antisymmetric form F (that is, one p-brane), so that Eqs. (5.133)–(5.136) yield the general static
solution to the field equations, Eqs. (5.675) contain various linear combinations of δxK with u-
dependent coefficients.
There is, however, an important case when Eqs. (5.675) do decouple for any configuration of M
with the metric (5.111), namely, the single-brane solution (5.133)–(5.136) under the condition that
the vector c = (cK ) is parallel to Z in V: 5
cK = BY K /Y 2 , B = const (5.678)
(here and henceforth in this section we omit the index s since, by our assumption, it takes only one
value). This condition is automatically valid for the case of utmost interest, BHs with one p-brane
(“a black p-brane”), which, by (5.143), corresponds to B = k = h ≥ 0.
Due to the collinearity condition (5.170) and the constraint (5.135), the constants are now
connected by the relation
N 0 (h2 sign h − B 2 ) = k 2 sign k − B 2 (5.679)
with N 0 = Y −2 (d0 − 1)/d0 < 1. It turns out that, besides BHs, the condition (5.678) is satisfied for
some singular solutions whose behaviour is quite generic for the system under study:
1. k > 0, h ≥ 0, such that umax = ∞ and a singularity at the centre of symmetry is attractive
at least in terms of the metric (5.665), eγ → 0;
2. h < 0, so that the solution behaviour is determined by the function s(h, u+u1 ) = h−1 sin h(u+
u1 ) in (5.133) where u1 = const ∈ (0, π/|h|). In this case the central singularity is repulsive,
eγ → ∞, of Reissner-Nordström type.
Due to (5.678), Eqs. (5.675) take the form
h 1 ³f2 ´ i y + Bu
u def
L̂δxK = 2Z K − fuu (Zδx), f (u) = − (5.680)
d0 βu u Y2
with y(u) determined by (5.133); the area function β has the form
1 n 1 o
β=− ln[(d0 − 1)s(k, u)] + 0 f (u) − Bu + const (5.681)
d0 − 1 N
where the value of the constant is inessential.
Since V is an l-dimensional Euclidean space (l = n−1+ |A|), there are l −1 linearly independent
vectors Z ⊥ such that Z ⊥ Z = 0. Therefore the set of wave equations (5.680) decouples into one
equation for Zδx and l − 1 equations for different Z ⊥ δx:
h 1 ³f2 ´ i
u
L̂ (Zδx) = U (u)(Zδx), U (u) = 2Z 2 − fuu ; (5.682)
d0 βu u
L̂ (Z ⊥ δx) = 0. (5.683)
5
Curiously, they are parallel in V although the corresponding vectors ~c and Y~ are mutually orthogonal in the
surrounding target space V. For a clear picture, imagine two vectors in 3-dimensional space whose projections onto
a plane lie on the same ray.
CBPF-MO-003/02 101
The static nature of the background solution makes it possible to separate the variables:
0
Zδx = ψ(u) eΩt , Z ⊥ δx = ψ 0 (u) eΩ t , (5.684)
The existence of an admissible solution of any of these equations with a real value of Ω or Ω0 would
mean that the perturbation can grow exponentially with time, hence the instability.
It is hard to solve Eqs. (5.685), (5.686) in their full range but it is rather easy to assess the
asymptotic behaviour of their solutions near u = 0 and umax , and this will be sufficient for making
stability conclusions.
In particular, for u → 0, which corresponds to spatial infinity, one has
The general asymptotic form of solutions to (5.685) and (5.686) at small u for all cases under study
may be written as follows:
h i
ψ (or ψ 0 ) = ud0 /(2d) c1 exp(c0 d Ωu−1/d ) + c2 exp(−c0 d Ωu−1/d ) , c1 , c2 = const. (5.688)
The boundary condition (5.676) then requires that in (5.688) c1 = 0, and it remains to look at the
other end of the u range, u → umax .
Instability of naked singularities
Consider Case 1 of the previous subsection, a “scalar type” singularity. As u → ∞, the relevant
functions of the static solution behave as follows:
u N1 − 1
y = hu + O(1), β≈− (k − B)(h − B) (5.689)
Y 2 d0 h + k
one sees that eβ(u) → 0 exponentially. The same happens to U (u), therefore the asymptotic form
0
of (5.685) or (5.686) is simply ψuu = 0 or ψuu = 0, so that
ψ (or ψ 0 ) = c3 u + c4 (5.690)
with constants c3 and c4 . On the other hand, the background functions xK also behave as const · u
as u → ∞, therefore the second boundary condition (5.677) is satisfied for any solution (5.690),
including the one joining the solution (5.688) with c1 = 0 at small u. We conclude that there are
growing modes of perturbations for any Ω, hence the singular solution is catastrophically unstable.
In Case 2, h < 0, we have umax = π/|h| − u1 < ∞ and the relevant functions in the static
solution approach umax in the following way:
ZK 1
y(u) ≈ − ln(|h|∆u), xK ≈ − ln(|h|∆u), β≈ ln ∆u → −∞ (5.691)
Y2 d0 Y 2
CBPF-MO-003/02 102
where ∆u = umax − u. One can make sure that U (u) does not affect the asymptotic behaviour of
solutions to Eq. (5.685) as u → umax as compared with that of Eq. (5.686), and for both one can
write:
ψ (or ψ 0 ) = c5 + c6 ∆u (5.692)
while the condition (5.677) only requires |δxK / ln ∆u| < ∞. Thus the solution satisfies (5.677) for
any choice of the constants c5 , c6 , and, as in Case 1, this leads to the instability of the background
singular solution.
and so there is no reason to require |δxK | < ∞. In some cases it is regular (see Sec.3.2 and examples
in 5.6.8). One can see, however, that again, as u → ∞, Eqs. (5.685) and (5.686) for a single-brane
extremal BH take the form ψuu = 0; the linearly growing solution is discarded since it grows faster
than xK in (5.694), so we are left with a constant and have to require |ψ < ∞| for both regular
and singular backgrounds. Then the same reasoning with ψuu /ψ > 0 makes us conclude that such
allowed solutions with Ω > 0 do not exist and extremal BHs are stable as well. Indeed, an explicit
form of U (u) is
2Z 2 h 2 00 2 i
0 −u1 + N u
U (u) = 2 1 + N (5.695)
Y (u + u1 )2 (u1 + N 00 u)2
where N 0 < 1 was defined in (5.693) and N 00 = 1 − N 0 . The reasoning works since U > 0 for all
u > 0 and U → 0 as u → ∞.
CBPF-MO-003/02 103
We can now formulate the following result, to be used in the further consideration:
Proposition 2. If a decoupled linear perturbation mode ξ of a static, spherically symmetricBH
solution obeys the equation
L̂ ξ = U (u)ξ (5.696)
with U (u) ≥ 0 (including the case U ≡ 0), this mode is stable.
with
1 h (ys,u + k)(ys0 ,u + k) i
fss0 (u) = . (5.698)
Ys2 Ys20 d0 βu u
Just as in the one-brane case, one easily separates the “transversal” degrees of freedom: for
vectors Z ⊥ ∈ V such that Z ⊥ Z s = 0 (they fill a (dim V − 2)-dimensional plane), the function
ξ = Z ⊥ δx obeys the wave equation (5.696) with U ≡ 0.
However, Eqs. (5.675) with the matrix (5.697) in general do not decouple. An exception is the
special case when the two functions ys coincide,
although the vectors Y~ s are different; in the BOS terminology (Sec. 3.3) the two vectors Y~ s form
a block and in our case this single block exhausts the whole system. If we suppose for simplicity
that the norms of Y~ s coincide, Y12 = Y22 = Y 2 , then the charges coincide as well, Q21 = Q22 = Q2 ,
and one obtains for the two modes ξ± = (oZ1 ± Z 2 )δx:
£ ¤
L̂ ξ+ = 2Z 2 (1 + cos θ) Q2 e2y + 2F ξ+ ; (5.700)
L̂ ξ− = 2Z 2 Q2 e2y (1 − cos θ)ξ− (5.701)
where Z 2 = Y 2 − (d0 − 1)/d0 (see (5.663), θ is the angle between the vectors Z 1 and Z 2 in V and
F is the function (5.698) with Y12 = Y22 and y1 = y2 = y.
A direct substitution of y and βu into (5.700) shows, as before, that the coefficient by δx+ at the
r.h.s. is nonnegative; for δx− in (5.701) this is manifestly so. Hence the previous reasoning works
and we conclude that such BHs (including extremal ones) with two branes are stable.
The case Y12 6= Y22 is covered in the next section.
CBPF-MO-003/02 104
It is easy to see that, due to (5.663), similar relations hold for the corresponding vectors Z s ∈ V:
X
Zω = as Z s ∀s; Z s Z ω = Z 2ω . (5.703)
s
For certainty we suppose that Z s are linearly independent; if they are not, the consideration is
slightly modified without changing the results.
The wave equations (5.675) take the form
1 X 1 h (yu + k)2 i
L̂ δxK = qb e2y(u) as Zs K ZsL δxL + F (u)Zω K Zω L δxL , F (u) = , (5.704)
2 s
Yω4 d0 βu u
P ~⊥
~ ⊥ δx (where Z
where qb = qbω = s Q2s and y(u) is given by (5.699). As before, the perturbations Z
belong to the plane V⊥ orthogonal to all Z~ s , whose dimension is dim V − m ≥ 0) are decoupled and
obey the equation (5.683), giving no unstable modes.
Multiplying (5.704) by Z ω , one obtains a decoupled equation for ξω = Z ω δx:
where Zω2 = Z 2ω . Since, as is directly verified, Uω (u) ≥ 0, this mode is also stable.
The remaining (m−1) degrees of freedom may be described in terms of the vectors W s = Z s −Z ω
and the functions ξs = W s δx, such that
X X
W s Z ω = 0; as W s = 0; as ξs = 0. (5.706)
s s
Using (5.703) and (5.706), one obtains the following m equations, coupled due to (5.706), for (m−1)
independent variables:
m
X
q e2y
L̂ ξs = 2b Kss0 ξs0 , Kss0 = as0 W s W s0 . (5.707)
s0 =1
Excluding one of the unknowns, say, ξm , by virtue of (5.706), we arrive at a determined set of wave
equations for ηs = ξs − ξm , s = 1, . . . , m − 1:
m−1
X ³ m−1
1 X ´
q e2y
L̂ ηs = 2b Fss0 ηs0 , Fss0 = as0 W s0 W s + as00 W s00 . (5.708)
s0 =1
am s00 =1
CBPF-MO-003/02 105
This is a good way of studying specific models. In the general case, however, the situation looks more
transparent if we consider, instead, an auxiliary system with m independent unknowns, described by
Eqs. (5.707) where W m is slightly shifted from its true value by some ∆W , so that all W s become
linearly independent; the relation among ξs in (5.706) is then cancelled as well. Our system is
restored when ∆W → 0.
For the auxiliary system the matrix (W s W s0 ) is symmetric and positive-definite; if all as are
equal, the same is true for the matrix of coefficients in (5.707), (Kss0 ), hence there is a similarity
transformation bringing it to a diagonal form with its positive eigenvalues along the diagonal. Such
a transformation applied to Eqs. (5.707) decouples them into m separate wave equations like (5.705),
with some positive function replacing U (u). In the limit ∆W → 0, the worst thing that can happen
is that some of the eigenvalues tend to zero, giving for some combinations of ξs the equation L̂ ξ = 0
which, as we know, does not lead to an instability. One can assert “by continuity” that this picture
is generic, at least for as close enough to being equal, and stability is again concluded according to
Proposition 2.
On the contrary, when the numbers am > 0 are different, one cannot guarantee that the non-
symmetric matrix Kss0 is similar to a diagonal one [351]. A failure in its diagonalization can be
connected with the occurrence of a pair (or pairs) of complex roots λs of the characteristic equation
det |Kss0 − λδss0 | = 0. In this case there is at least one pair of coupled perturbations for which
a special investigation is necessary. An inspection of the characteristic equation shows that the
matrix Kss0 cannot have negative eigenvalues, therefore a separate unstable mode cannot occur and
the only possible instability can be connected with coupling between modes.
In particular, in an arbitrary OS BH solution there is a subfamily where all ys (u) coincide (i.e.,
the constants us are the same for all s), so that the branes form a BOS block, and it turns out
that all as are also equal, as well as the squared charges Q2s . The above reasoning shows that such
solutions are stable.
If rank(Kss0 ) < m − 1, that is, there are additional linear dependences among Z s , then some
combinations of ξs decouple leading to equations of the form L̂ ξ = 0, and for the remaining modes
the above discussion can be repeated with slight modifications.
This is what can be said about the general case of single-block BOS BH solutions. If there is
a block of only two branes (m = 2), one can make a common stability conclusion generalizing the
one made in earlier. Indeed, for ξω = a1 ξ1 + a2 ξ2 there is Eq. (5.705), whereas for ξ− = ξ1 − ξ2 one
obtains
q e2y a1 (1 − a1 )(Z 1 − Z 2 )2 ξ− .
L̂ ξ− = 2b (5.709)
In the special case Z12 = Z22 = Z 2 one recovers (5.700), (5.701).
For m ≥ 3 one has to study specific models individually.
ys coincide). It turns out that for such backgrounds the wave equations for perturbations also
generically decouple and the absence of unstable modes can be proved. Though, such a general
proof is available only in two cases: (i) two-brane BH solutions (m = 2) and (ii) equal-charge
subfamilies of arbitrary OS solutions. Nontrivial brane systems with m > 2 should be studied
individually to see whether or not the corresponding matrix (Kss0 ) in (5.707) can be diagonalized
over the field of real numbers. If yes, the solution is stable, otherwise a further study of coupled
modes is necessary.
It should be stressed that a BOS-block solution exists for an arbitrary set of linearly independent
vectors Y~ s . In particular, if any multi-brane static BH solution for a certain set of input parameters
with independent vectors Y~ s is known, e.g., any OS or BOS solution (see 5.6.3), then the additional
requirement that all the functions ys coincide selects from it a special BOS-block solution, for which
a stability study can be performed as described above. The only restriction is the requirement
aµ (ω) > 0 for the charge factors obeying Eqs. (5.139).
Some technical points are worth mentioning. First, in gravitational stability studies it is some-
times rather hard to separate real physical perturbations from purely gauge degrees of freedom. We
avoid this problem by obtaining the set of wave equations (5.675) where the number of equations
is precisely the number of dynamical degrees of freedom, represented by scalars in the physical
space-time M phys . Due to the latter circumstance, one more complication is avoided: when dealing
with vector and tensor perturbations of BHs, one has to take into account the apparent singularity
of the metric on the horizon; to properly formulate the boundary conditions, it is then necessary to
pass to Kruskal-like coordinates; to be admissible, and the perturbations are required to be finite
on the future horizon [339, 345, 346]. In our case the perturbations are scalars, so the finiteness
requirement can be imposed in any coordinates. The choice of gauge only remains important for
making the treatment more transparent.
To conclude, we would like to emphasize that our consideration did not depend on the number
and dimensions of the factor spaces in the original space-time M , on the number of scalar fields ϕa
and on the particular values of their coupling constants λsa .
5.6.8 Examples
Consider, for illustration, some solutions of D = 11 supergravity, representing the low-energy limit
of M-theory, as examples of systems to which our stability results apply.
The action (5.110) for this theory does not contain scalar fields (ϕa = λsa = 0) and the only F -
form is of rank 4, whose various nontrivial components Fs (elementary F -forms according to 5.6.2, to
be called simply F -forms) are associated with electric 2-branes [for which d(Is ) = 3] and magnetic
5-branes [such that d(Is ) = 6] (see [336] and references therein). The orthogonality conditions
(5.132) are satisfied if the following intersection rules hold:
3 ∩ 3 = 1, 3 ∩ 6 = 2, 6 ∩ 6 = 4. (5.710)
consist of a single term. These solutions are well known; the metric (5.111) can be presented as
h 1 − 2k/(drd ) ³ dr2 ´ i
ds211 = H d(I)/9 − dt2 + + r2 dΩ2 + H −1 ds2on + ds2off (5.711)
H 1 − 2k/(drd )
p
where H = H(r) = 1 + p/(drd ), p = k 2 + 2Q2 − k, d = d0 − 1; ds2on and ds2off are the “on-brane”
and “off-brane” extra-dimension line elements, respectively; the dimension d0 of the sphere M 0
varies from 2 to 7 for d(I) = 3 (an electric brane) and from 2 to 4 for d(I) = 6 (a magnetic brane).
In particular, the cases of maximum d0 , when off-brane extra dimensions are absent, correspond
in the extremal near-horizon limits to the famous structures AdS4 × S 7 (electric) and AdS7 × S 4
(magnetic). All these solutions are stable under linear spherically symmetricperturbations.
2. Some examples of orthogonal systems (OS), whose stability in the general case is yet to be
studied, are:
(i) (123), (145), (167) — 3 electric branes; d0 = 2 or 3.
(ii) (123), (124567) — 1 electric and 1 magnetic branes; d0 = 2 or 3.
(iii) (123), (145), (124678), (135678); d0 = 2.
OS
The metrics are easily found from (5.145) with D − 2 = 9, Ys2 = 2 and the equalities marked = .
The systems (i) with d0 = 3 and (iii) are remarkable in that their extremal limits have regular
horizons and the near-horizon geometries are, respectively, AdS2 × S 3 × T 6 and AdS2 × S 2 × T 7 if
the remaining extra dimensions are compactified on tori.
When the orthogonal systems form BOS-blocks (i.e., in the special case of equal charges and a
unique function y(u)), the solutions are stable according to 5.6.6.
3. All two-brane BOS-block BHs are stable according to 5.6.6, for instance,
(i) (123), (123456); Y~ 1 Y~ 2 = 1. (ii) (123), (145678); Y~ 1 Y~ 2 = −1.
The norms are equal (Ys2 = 2), and the angle θ is 60◦ in case (i) and 120◦ in case (ii).
4. Many seemingly possible three-brane blocks turn out to be forbidden due to a zero value of a
certain as (see 5.6.3). Consider, e.g.,
2 0 1
(123), (145), (123678); (Y~ s Y~ s0 ) = 0 2 −1 . (5.712)
1 −1 2
One easily finds from (5.139) that (a1 , a2 , a3 ) = (0, 1/2, 1/2), so one of the charges should be zero,
which means that such a system cannot exist.
5. The following is an example of a single-block BH whose stability can be established by an
individual study as described in 5.6.6:
2 0 1
(123), (145), (123467); d0 = 2 or 3; (Y~ s Y~ s0 ) = 0 2 0. (5.713)
1 0 2
From (5.139) it follows
3
X
(a1 , a2 , a3 ) = (2/7, 3/7, 2/7); Y~ ω = as Y~ s ; Y~ 2ω = 6/7. (5.714)
s=1
CBPF-MO-003/02 108
The next stage is to separate the perturbation Z ω δx, after which the remaining two degrees of
freedom obey Eqs. (5.708) of the form
2
X µ ¶
2y 2/7 0
L̂ ηs = 2b
qe Fss0 ηs0 , Fss0 = . (5.716)
−2/7 6/7
s0 =1
The characteristic equation det |Fss0 − λδss0 | = 0 has the form (2/7 − λ)(6/7 − λ) = 0, and, according
to Proposition 2, the positivity of its roots proves the stability of the background configuration.
which is unobservable directly since our brane world is located in a potential “trench” and/or most
of the types of matter are concentrated on this brane. The recent boom was apparently launched by
the works of Randall and Sundrum [358] who showed, in particular, a way of obtaining Newtonian
gravity on the brane from a multidimensional model. Since their publication hundreds of papers
have appeared, with a diversity of specific models and predictions. We will not try here to review
this vast trend since it seems too early to make conclusions: while one is preparing a survey, tens
of new works, drastically changing the picture, can appear. We would only mention an opportunity
suggested by Maia and Silveira [359] well before the present outburst. These authors argued that
near a black hole (BH) particles may gain sufficient gravitational energy to overcome the potential
barrier confining them to four dimensions and can thus run away from our world.
In this paper we discuss in some detail predictions from extra dimensions connected with lo-
cal sources of gravity. Assuming that extra dimensions of variable size inside and around such
sources (e.g., stars, galaxies, black holes) can affect various physical phenomena, one should apply
multidimensional theories to describe these phenomena.
In Sec. 5.7.2 and 5.7.3 we present some exact static, spherically symmetric solutions of a gen-
eralized field model [105,143,147] associated with charged p-branes and motivated by the bosonic
sector of the low-energy field approximation of superstring theories, M-theory and their generaliza-
tions [31,6,102]. Our model, however, is not restricted to known theories since it assumes arbitrary
dimensions of factor spaces, arbitrary ranks of antisymmetric forms and an arbitrary number of
scalar fields.
Then, in Sec. 5.7.4, we discuss the choice of the conformal frame (CF) in which observational
predictions should be formulated. Since at present there is no generally accepted unified theory, in
our approach we use a generalized model with arbitrary p-branes in diverse dimensions and study
the physical applications on the basis of exactly solvable models. Thus we do not fix the underlying
fundamental theory and have no reason to prescribe a particular frame, therefore all further results
are formulated in an arbitrary frame.
In Sec. 5.7.5 we derive the post-Newtonian (PN) approximation of the above solutions in order to
designate possible traces of extra dimensions and p-branes in the comparatively weak gravitational
fields of the great majority of planetary and stellar systems, including binary pulsars. This section
generalizes the results of previous papers [223,248]. For black holes, apart from the PN parameters
which determine the motion of test bodies in their sufficiently far neighbourhood, there is one more
potentially observable parameter, the Hawking temperature TH , which is obviously important for
small (e.g., primordial) black holes rather than those of stellar or galactic mass range. We discuss the
expressions for TH for a variety of black hole solutions and their applicability in different conformal
frames.
Sec. 5.7.6 and 5.7.7 describe such consequences of these (and many other) field models as the
Coulomb law violation and the possible existence of new, purely multidimensional objects, T -holes
[362, 363]. The results are briefly discussed in Sec. 5.7.8.
The model
The starting point is, as in Refs. [143, 105, 147] the model action for D-dimensional gravity
with several scalar dilatonic fields ϕa and antisymmetric ns -forms Fs :
Z p n X ηs o
1 a
S = 2 dD z |g| R[g] − δab g M N ∂M ϕa ∂N ϕb − e2λsa ϕ Fs2 , (5.717)
2κ s∈S
ns !
M
CBPF-MO-003/02 110
Here u is a radial coordinate ranging in Ru ⊆ R; ds20 = dΩ2 is the metric on a unit d0 -dimensional
sphere M 0 = Sd0 ; t ∈ M 1 ≡ Rt is time; the metrics g i = ds2i of the “extra” factor spaces (i ≥
2) are assumed to be u-independent, Ricci-flat and can have arbitrary signatures εi = sign g i ;
M ...M
|g| = | det gM N | and similarly for subspaces; Fs2 = Fs, M1 ...Mns Fs 1 ns ; λsa are coupling constants;
ηs = ±1 (to be specified later); s ∈ S, a ∈ A, where S and A are some finite sets.
This formulation admits both spacelike and timelike extra dimensions. Models with multiple
timelike dimensions were considered in a number of papers, e.g., [102] and more recently in [360, 361].
i
The “scale factors” eβ and the scalars ϕa are assumed to depend on u only.
The F -forms should also be compatible with spherical symmetry. A given F -form may have
several essentially (non-permutatively) different components; such a situation is sometimes called
“composite p-branes”6 . For convenience, we will nevertheless treat essentially different components
of the same F -form as individual (“elementary”) F -forms. A reformulation to the composite ansatz,
if needed, is straightforward.
The ns -forms F = F[ns ] = dAs , where As is a potential (ns − 1)-form, are naturally classified
as electric (F eI ) and magnetic (FmI ) ones, both associated with a certain subset I = {i1 , . . . , ik }
(i1 < . . . < ik ) of the set of numbers labeling the factor spaces: {i} = I0 = {0, . . . , n}. By definition,
the potential AI of an electric form F eI carries the coordinate indices of the subspaces M i , i ∈ I and
is u-dependent (since only a radial component of the field may be nonzero), whereas a magnetic form
FmI is built as a form dual to a possible electric one associated with I. Thus nonzero components
def
of FmI carry coordinate indices of the subspaces M i , i ∈ I = I0 \ I. One can write:
n eI = rank F eI = d(I) + 1, nmI = rank FmI = D − rank F eI = d(I) (5.719)
P
where d(I) = i∈I di are the dimensions of the subspaces M I = M i1 × . . . × M ik . The index s will
be used to jointly describe the two types of forms, so that [155,214]
S = {s} = { eIs } ∪ {mIs }. (5.720)
We will make some more natural assumptions:
(i) The branes only “live” in extra dimensions, i.e., 0 6∈ Is , ∀s.
(ii) The energy density of each F -form is positive: −Ttt (Fs ) > 0, ∀s.
When all extra dimensions are spacelike, the second requirement holds if, as usual, in (5.209)
ηs = 1 for all s. In more general models, with arbitrary εi , the requirement −Ttt > 0 holds if
η eI = −ε(I)εt (I), ηmI = −ε(I)εt (I), (5.721)
Y ½
def 1, Rt ⊂ M I ,
ε(I) = εi , εt (I) = (5.722)
−1 otherwise
i∈I
6
There is an exception: two components, having only one noncoinciding index, cannot coexist since in this case
N
there emerge nonzero off-block-diagonal components of the energy-momentum tensor (EMT) TM , while the Einstein
tensor in the l.h.s. of the Einstein equations is block-diagonal. See more details in Ref. [147]
CBPF-MO-003/02 111
where Rt is the time axis. If εt (I) = 1, we are dealing with a true electric or magnetic form,
directly generalizing the familiar Maxwell field; otherwise the F -form behaves as an effective scalar
or pseudoscalar in the physical subspace. F -forms with εt (I) = −1 will be called quasiscalar.
Several electric and/or magnetic forms (with maybe different coupling constants λsa ) can be
associated with the same I and are then labelled by different values of s. (We sometimes omit the
index s by I when this cannot cause confusion.)
The forms Fs are associated with p-branes as extended sources of the spherically symmetric field
distributions, where the brane dimension is p = d(Is ) − 1, while d(Is ) is the brane worldvolume
dimension.
The following example illustrates the possible kinds of F -forms.
Example 1: D = 11 supergravity, representing the low-energy limit of M-theory [6]. The
action (5.209) for the bosonic sector of this theory (truncated by omitting the Chern-Simons term)
does not contain scalar fields (ϕa = λsa = 0) and the only F -form is of rank 4, whose various
nontrivial components Fs (elementary F -forms, to be called simply F -forms according to the above
convention) are associated with electric 2-branes [for which d(Is ) = 3] and magnetic 5-branes [such
that d(Is ) = 6, see (5.719)]. The action has the form
Z p h
1 1 2i
S = 2 d11 z |g| R[g] − F[4] . (5.723)
2κ 4!
M
Let us put d0 = 2 and ascribe to the external space-time coordinates the indices M = u, θ, φ, t
(θ and φ are the spherical angles), and let M = 2, . . . , 8 refer to the extra dimensions. Furthermore,
let the extra factor spaces M i , i = 2, . . . , 8, be one-dimensional and coincide with the respective
coordinate axes. The number i = 1 is ascribed to the time axis, M 1 = Rt , as stated previously.
Then different kinds of forms can be exemplified as follows:
Fut23 is a true electric form, I = {123}; I = {045678}.
Fθφ23 is a true magnetic form, I = {145678}; I = {023}.
Fu234 is an electric quasiscalar form, I = {234}; I = {015678}.
Fθφt2 is a magnetic quasiscalar form, I = {345678}; I = {012}.
Q
where λs ϕ = λsa ϕa , ε... and ε... are Levi-Civita symbols, |gI | = i∈I |g i |, and Qs = const are charges.
In a similar way, for a magnetic m-form Fs (s = mI, m = d(I s )), the field equations and the Bianchi
identities dFs = 0 lead to
p 1 2
Fs,M1 ...Md(I) = Qs εM1 ...Md(I) |gI | ⇒ F = ε(I)Q2s e−2σ(I)+2λs ϕ (5.727)
m! s
We use the notations n
X X
σi = dj β j (u), σ(I) = di β i (u). (5.728)
j=i i∈I
where δiI = 1 if i ∈ I and δiI = 0 otherwise. The contravariant components and scalar products of
the vectors Y~ s are found using the matrix GAB inverse to GAB :
µ ij ¶ ³ ´
AB δ /di − 1/(D − 2) 0 A d(I)
(G ) = , (Y s ) = δiI − , −χs sa ;
λ
0 δ ab D−2
(5.738)
d(Is )d(Is0 )
Ys,A Ys0 A ≡ Y~ s Y~ s0 = d(Is ∩ Is0 ) − + χs χs0 λs λs0 . (5.739)
D−2
The equations of motion in terms of Y~ s read
X def
ẍA = qs Ys A e2ys , qs = ²s Q2s . (5.740)
s
where the vectors of integration constants ~c and ~c are orthogonal to all Y~ s : cA Ys,A = cA Ys,A = 0,
or
ci di δiIs − ca λsa = 0, ci di δiIs − ca λsa = 0. (5.745)
such that the vectors Y~ µ(ω) (µ(ω) ∈ Sω ) form mutually orthogonal subspaces Vω ⊆ V:
def
Y~ µ Y~ ω = Yω2 = Y~ 2ω , ∀ µ ∈ Sω . (5.748)
Then one has the following solution to the equations of motion (5.740), (5.735):
X Yω A
xA = yω (u) + cA u + cA , (5.749)
Yω2
½ω X
qbω Yω2 s2 (hω , u + uω ), qbω > 0, def
e−2yω = 2 qbω = ²s Q2µ , (5.750)
|bq ω |Yω2 h−2
ω cosh [h ω (u + u ω )], qbω < 0, hω > 0;
ω
X h2 sign hω d0
ω
E = 2
+ ~c 2 = k 2 sign k, (5.751)
ω
Yω d0 − 1
where hω , uω , cA and cA are integration constants; cA and cA are constrained by the orthogonality
relations (5.745) (the vectors ~c and ~c are orthogonal to each individual vector Y~ s ∈ V); the function
s(., .) has been defined in (5.224).
Eqs. (5.748) form a set of linear algebraic
P equations with respect to the “charge factors” aν =
2
²ν Qν /b
q ω 6= 0, satisfying the condition ω aµ = 1. The existence of a solution to (5.748) guarantees
that qbω 6= 0. On the other hand, if a solution to (5.748) gives aµ = 0 for some µ ∈ Sω , this means
that the block cannot contain such a p-brane, and then the consideration may be repeated without
it.7
7 ~ ω solving Eqs. (5.748) is the altitude of the pyramid formed by the vectors Y
Geometrically, the vector Y ~ µ , µ ∈ Sω
with a common origin. The condition aµ > 0 means that this altitude is located inside the pyramid, while aµ = 0
means that the altitude belongs to one of its faces.
CBPF-MO-003/02 115
The function yω (u) is equal to yµ(ω) (u) = Yµ(ω),A xA , which is, due to (5.748), the same for all
µ ∈ Sω . The BOS solution generalizes the OS one, (5.234), (5.235): the latter is restored when each
block contains a single F -form.
Both kinds of solutions are asymptotically flat, and it is natural to normalize the functions ys (u)
and yω (u) by the condition ys (0) = 0 or yω (0) = 0, so that the constants us and uω are directly
related to the charges.
Other solutions to the equations of motion are known, connected with Toda chains and Lie
algebras [163, 3, 238].
Black-hole solutions
Black holes (BHs) are distinguished among other spherically symmetric solutions by the exis-
tence of horizons instead of singularities in the physical 4-dimensional space-time M phys ; the extra
dimensions and scalar fields are also required to be well-behaved on the horizon to provide regu-
larity of the D-dimensional metric. Thus BHs are described by the above solutions under certain
constraints upon the input and integration constants. The no-hair theorem of Ref. [228] states that
BHs are incompatible with quasiscalar F -forms. This means that all ²s = 1, hence, in particular,
in the above BOS solution (5.748)–(5.751),
p qbω > 0 and all aν > 0. Furthermore, requiring that all
βi β1
the scale factors e (except e = |gtt | which should tend to zero) and scalars ϕa tend to finite
limits as u → umax , we get [214]:
X
hω = k > 0, ∀ ω; cA = k Yω−2 Yω A − kδ1A (5.752)
ω
where A = 1 corresponds to i = 1 (time). The constraint (5.744) then holds automatically. The
value u = umax = ∞ corresponds to the horizon. The same condition for the OS solution (5.234)–
(5.745) is obtained by replacing ω 7→ s.
Under the asymptotic conditions ϕa → 0, β i → 0 as u → 0, after the transformation
2k def
e−2ku = 1 − , d = d0 − 1 (5.753)
drd
the metric (5.225) for BHs and the corresponding scalar fields may be written as
³Y ´· ³ 2k ´ Y −2/Yω2
2 Aω 2
dsD = Hω −dt 1 − Hω
ω
drd ω
³ ´ X n Y ¸
dr2 2 2 2 Aiω
+ + r dΩ + dsi Hω ;
1 − 2k/(drd ) i=2 ω
def 2 X OS 2
Aiω = − 2 aµ δiIµ = − δiI ; (5.754)
Yω µ∈S Ys2 s
ω
X 1 X OS
X λsa
ϕa = − 2
ln H ω aµ λµa = − ln Hs , (5.755)
ω
Y ω µ∈S s
Ys2
ω
OS
where = means “equal for OS, with ω 7→ s”, and Hω are harmonic functions in R+ × Sd0 :
def p
Hω (r) = 1 + Pω /(drd ), Pω = k 2 + qbω Yω2 − k. (5.756)
CBPF-MO-003/02 116
The subfamily (5.244), (5.246)–(5.247) exhausts all BOS BH solutions with k > 0; the OS
ones are obtained in the special case of each block Sω consisting of a single element s. The only
independent integration constants remaining in BH solutions k, related to the observed mass (see
below), and the brane charges Qs .
Example 2. The simplest, single-brane BH solutions are described by (5.246), (5.247) where
all sums and products in s consist of a single term. These solutions are well known [364]. the metric
(5.210) for, e.g., D = 11 supergravity (5.723) can be presented as
h 1 − 2k/(drd ) ³ dr2 ´ i
ds211 = H d(I)/9 − dt2 + + r2 dΩ2 + H −1 ds2on + ds2off (5.757)
H 1 − 2k/(drd )
p
where H = H(r) = 1 + P/(drd ), P = k 2 + 2Q2 − k, d = d0 − 1; ds2on and ds2off are r-independent
“on-brane” and “off-brane” extra-dimension line elements, respectively; the dimension d0 of the
sphere M 0 varies from 2 to 7 for d(I) = 3 (an electric brane) and from 2 to 4 for d(I) = 6 (a
magnetic brane). In particular, the cases of maximum d0 , when off-brane extra dimensions are
absent, correspond in the extremal near-horizon limits to the famous structures AdS4 × S7 (electric)
and AdS7 × S4 (magnetic). All these BH solutions are stable under linear spherically symmetric
perturbations [352]; though, small multidimensional BHs, whose horizon size is of the order of the
conpactification length, are known to possess the Gregory-Laflamme instability [346, 347] related
to distortions in extra dimensions.
The above relations describe non-extremal BHs. Extremal ones, corresponding to minimum
BH mass for given charges (the so-called BPS limit), are obtained in the limit k → 0. The same
solutions follow directly from (5.751)–(5.750) under the conditions hω = k = cA = 0. For k = 0,
the solution is defined in the whole range r > 0, while r = 0 in many cases corresponds to a naked
singularity rather than an event horizon, so that we no more deal with a black hole. However, in
many other important cases r = 0 is an event horizon of extremal Reissner-Nordström type, with
an AdS near-horizon geometry and even the global metric turns out to be regular, as it happens for
the AdS4 × S7 and AdS7 × S4 structures mentioned in the previous paragraph [6];
Other families of solutions, mentioned at the end of the previous section, also contain BH
subfamilies. The most general BH solutions are considered in Ref. [238].
where gµν is the 4-dimensional part of the original metric gM N used in (5.209), while eσ2 , defined
in (5.728), is the volume factor of all extra dimensions.
The choice of a physical CF in non-Einsteinian theories of gravity is widely discussed, but the
discussion is mostly restricted to the 4-dimensional metric — see e.g. [365, 366, 367] and numerous
references therein. There are arguments in favour of the Einstein frame, and the most important
ones, applicable to higher order and scalar-tensor theories (and many multiscalar-tensor theories
obtainable from multidimensional gravity) are connected with the positivity of scalar field energy
and the existence of a classically stable ground state [365, 366]; though, these requirements are
violated if quantum effects are taken into account [4].
In our view, however, the above arguments could be convincing if we were dealing with an “ab-
solute”, or “ultimate” theory of gravity. If, on the contrary, the gravitational action is obtained
as a certain limit of a more fundamental unification theory, theoretical requirements like the exis-
tence of a stable ground state should be addressed to this underlying theory rather than its visible
manifestation. In the latter, the notion of a physical CF should be only related to the properties of
instruments used for measuring lengths and time intervals. Moreover, different sets of instruments
(different measurement systems [4]) are described, in general, by different CFs.
Therefore, for any specific underlying theory that leads to the action (5.209) in a weak field limit,
two CFs are physically distinguished: one, which may be called the fundamental frame, where the
theory is originally formulated and another one, the observational frame, or the atomic system of
measurements (the 4-A frame), providing the validity of the weak equivalence principle (or geodesic
motion) for ordinary matter in 4 dimensions. The fundamental frame is specified in the original
space-time where the theory is formulated and is a natural framework for discussing such issues as
space-time singularities, horizons, topology, etc. (what happens as a matter of fact). On the other
hand, the 4-A frame is necessary for formulating observational predictions (what we see), and its
choice depends on how fermions are introduced in the underlying theory [4,363]. The reason is that
as long as clocks and other instruments used in observations and measurements consist of fermionic
matter, the basic atomic constants are invariable in space and time by definition. For instance, the
modern definition of reference length is connected with a certain spectral line, determined essentially
by the Rydberg constant and, basically, by the electron and nucleon masses.
The (4-dimensional section of the) fundamental frame and the 4-A frame are, generally speak-
ing, different, and none of them necessarily coincides with the 4-E frame, which represents the
gravitational system of measurements [4].
If the underlying theory is string theory, the fundamental frame is realized by the so-called
“string metric” (see e.g., [369, 290]), connected with gM N of Eq. (5.209) (the D-E metric) by a
dilaton-dependent conformal factor. On the other hand, to distinguish the observational frame, we
have to take into account that even for a fixed underlying theory, such frames may be different
for different particular cosmological models. Thus, for the case of string theory, new results on
the equivalence of quantum and some classical dilatonic brane-worlds in string and Einstein frames
have been obtained in Ref. [368]. For a more general context of string theory, let us recall that,
in the effective field-theoretic limit of string theory in 10 dimensions, the Lagrangian is presented
in a form similar to (5.209) with some quadratic fermion terms do not contain the dilaton field
([31], Eq. (13.1.49)). If those terms are associated with matter, then, by analogy, it is reasonable
for illustration purposes to write the matter Lagrangian Lm in our generalized model simply as an
additional term in the brackets of Eq. (5.209).
CBPF-MO-003/02 118
The first restriction is a result of over 2 million VLBI observations [370]. The second one follows
from the γ data and an analysis of lunar laser ranging data. In this case a high precision test based
on the calculation of the combination (4β − γ − 3), appearing in the Nordtvedt effect [371], is used
[321].
For the multidimensional theory under consideration, the metric (5.760) should be identified
with the asymptotics of the 4-dimensional metric from a solution in the observational (4-A) frame.
Preserving its choice yet undetermined, we can write according to ([5.225]) with d0 = 2:
n e−2σ1 h du2 io
2β 1 2
ds∗4 = e 2f (u)
− e dt + 2 + dΩ2
(5.763)
s (k, u) s2 (k, u)
where f (u) is an arbitrary function of u, normalized for convenience to f (0) = 0. Recall that by our
notations σ1 = β 1 + σ2 , the function s(k, u) is defined in Eq. (5.224), and spatial infinity takes place
at u = 0. The choice of the frame (5.759) means f = σ2 /4. The 4-E frame (5.758) corresponds to
f = σ2 /2.
Passing to isotropic coordinates in (5.763) with the relations
dρ du du2 2 1
=− , + dΩ = (dρ2 + ρ2 dΩ2 ), (5.764)
ρ s(k, u) s2 (k, u) ρ2
CBPF-MO-003/02 119
1 h u2 i
= u 1 − k 2 sign k + O(u4 ) ,
ρ 4
so that u = 1/ρ up to cubic terms, and the decomposition in powers of 1/ρ up to O(ρ−2 ), needed
for comparison with (5.760), precisely coincides with the u-decomposition near u = 0.
Using this circumstance, it is easy to obtain for the mass and the Eddington parameters corre-
sponding to (5.763):
00
0 1 β 1 + f 00 2f 0 − σ20
GM = −β 1 − f 0 ; β =1+ , γ =1+ , (5.765)
2 (GM )2 GM
¯
0 ¯
where f = df /du¯ and similarly for other functions. The expressions (5.765) are quite general,
u=0
being applicable to asymptotically flat, static, spherically symmetric solution of any theory where the
EMT has the property Tuu + Tθθ = 0, which leads to the metric (5.225). They apply, in particular, to
all solutions of the theory (5.209) under the conditions specified, both mentioned and not mentioned
above and those yet to be found.
In the observational frame (5.759) we have
00
1 0 3β 1 + σ200 σ20
GM = − (3β 1 + σ20 ); β =1+ , γ =1− . (5.766)
4 8(GM )2 2GM
In particular, if f = N σ2 ,
½ X h ¾
1 2 (2N − 1)d(Is ) i 2ys (0) X
2 (1 − 2N )d(Is ) 2ys (0)
β−1= Q 1−N + e + Qs e ;
2(GM )2 s:² =+1 s D−2 s:² =−1
D−2
s s
(5.769)
recall that ²s = 1 refers to true electric and magnetic forms, ²s = −1 to quasiscalar ones. For
N = 1/2 and N = 1/4 we obtain the values of β for the frames (5.758) and (5.759), respectively.
Explicit expressions for M and γ (in frames other than 4-E) require the asymptotic form of the
solutions.
It is convenient, without loss of generality, to normalize the scale factors at spatial infinity in
such a way that eβi (0) = 1, i = 1, . . . , n, so that the real scales of the extra dimensions are hidden in
CBPF-MO-003/02 120
the factor space metrics g i . In a similar way, one can re-define the dilatonic fields: ϕa − ϕa (0) 7→ ϕa ,
so that ϕa (0) = 0, while the former asymptotic values of ϕa have been actually absorbed in the
charges Qs . Then all ys (0) = 0.
For all OS and BOS solutions it then follows that the constants ci are zero; the constants uω ,
hω and qbω in (5.750) are related by
(
qbω Yω2 s2 (hω , uω ), qbω > 0,
1= 2 −2 2
(5.770)
|bq ω |Yω hω cosh (hω uω ), qbω < 0, hω > 0.
In (5.771), the second line corresponds to a quasiscalar form Fs , while in (5.770) the second line
means that the summed squared charge qbω of the block Sω is dominated by quasiscalar forms.
In what follows we will only give expressions for BOS solutions; their OS versions are then
evident.
Eq. (5.770) gives
(
¯
dyω ¯ −(bq ω Yω2 + h2ω sign hω )1/2 , qbω > 0;
0
yω = ¯ = (5.772)
du u=0 ±(h2ω − |b q ω |Yω2 )1/2 , qbω < 0,
For OS qbω is replaced by ²s Q2s , the two lines refering to ²s = 1 and ²s = −1, respectively.
The quantities needed for calculating M and γ are
X n
A0 1X X
x = Yω−2 YωA yω0 A
+c , σ10 = Aω yω0 + di ci , (5.773)
ω
2 ω i=1
1 1 1
with AP ω defined in (5.246). The quantity σ2 can be obtained as σ1 − β ; as before, x = β ,
Yω1 = µ∈Sω aµ Yµ1 and Yµ1 = δ1Iµ − d(Iµ )/(D − 2). The values of M and γ are now easily found
from (5.765) in terms of the solution parameters for any given f of the above general form, f = F~ ~x.
In particular, for BH solutions (5.244)–(5.248) there is no need to change the coordinates from
u to r or ρ: it is sufficient to use Eqs. (5.244) for the constants. Moreover, BH solutions contain
only true electric and magnetic forms, ²s = +1 [228] and qbω > 0.
0
Thus, for instance, for BOS BH solutions the quantities β 1 and σ20 are
0
X 1 − bω X 1 − 2bω
β 1 = −k − Pω , σ20 = − (5.774)
ω
Yω2 ω
Yω2
def P
with Pω defined in (5.248) and bω = µ∈Sω aµ d(Iµ )/(D − 2). In the OS case bω becomes bs =
d(Is )/(D − 2).
Accordingly, for BHs in CFs with f = N σ2 we obtain
X Pω 1 − 2N X Pω
GM = k + 2
[1 − bω + N (1 − 2bω )], γ= (1 − 2bω ). (5.775)
ω
Yω GM ω Yω2
• The expressions for β depend on the input constants D, d(Is ) (hence on p-brane dimensions:
ps = d(Is ) − 1), on the mass M and on the charges Qs . For given M , they are independent of
other integration constants, emerging in the solution of the Toda system (5.740), and also on
p-brane intersection dimensions, since they are obtained directly from Eqs. (5.740) [?]. This
means, in particular, that β is the same for BH and non-BH configurations with the same set
of input parameters, mass and charges.
• According to (5.769), all p-branes give positive contributions to β in both frames (5.758) and
(5.759), therefore (5.769) combined with (5.761) leads to a general restriction on the charges
Qs for given mass and input parameters.
• The expressions for γ depend, in general, on the integration constants hs or hω and ci emerging
from solving Eqs. (5.740). For BH solution these constants are expressed in terms of k and
the input parameters, so both β and γ depend on the mass, charges and input parameters.
• In the 4-E frame, one always has γ = 1. The same is true for some BH solutions in all frames
with f = N σ2 . Indeed, a pair of electric and magnetic p-branes with equal |Qs |, corresponding
to F -forms F1 and F2 of equal rank (in particular, if F1 and F2 are the electric and magnetic
components of the same composite F -form), always forms a BOS block, with a1 = a2 = 1/2,
so that bω = 1/2, and this pair does not contribute to σ20 in (5.774). Evidently γ = 1 as
well for a BOS black hole containing several such dyonic pairs and no other F -forms. This
property was noticed in Ref. [3] for the frame f = 0.
BH temperature. BHs are, like nothing else, strong-field gravitational objects, while the PN
parameters only describe their far neighbourhood. An important characteristic of their strong-field
behaviour, potentially observable and depending on their multidimensional structure, is the Hawking
temperature TH . As with other observables, it is of importance to know the role of conformal frames
for its calculation. One can ascertain, however, that this quantity is CF-independent, at least if
conformal factors that connect different frames are regular on the horizon.
Indeed, if, in an arbitrary static, spherically symmetric space-time with the metric
the sphere r = rhor is an event horizon, its Hawking temperature can be calculated as
1
TH = lim eΓ−A |dΓ/dr| (5.777)
2πkB r→rhor
where kB is the Bolzmann constant. This expression [155], which is invariant under reparametriza-
tions of the radial coordinate r, is easily obtained from standard ones [372]. The factor eΓ−A is
insensitive to conformal transformations g 7→ e2f (r) g, whereas Γ is replaced by Γ + f . At a horizon,
Γ → −∞, and, if rhor is finite (such a coordinate always exists), |dΓ/dr| → ∞. Therefore TH
calculated according to Eq. (5.777) will be the same in all frames with different f (r) provided df /dr
is finite at r = rhor .
This is precisely the case with the BH metric (5.246) and any f formed as a linear combination
of β i (i > 1 and ϕa . Using the recipe (5.777), one obtains [214]
1 Y³ 2k ´1/Yω 1 Y³ 2k ´1/Ys
2 2
OS
TH = = . (5.778)
8πkkB ω 2k + Pω 8πkkB s 2k + Ps
CBPF-MO-003/02 122
up to higher order terms in qbω /(GM )2 [bω was defined in (5.774)]. This expression should be
substituted into (5.778). For larger charges the corresponding expressions are more involved.
The temperature of extremal BHs (k = 0) only depends on the charges and the input parameters
Yω . Moreover, by (5.778), for some sets of Yω2 , TH can tend to infinity as k → 0. This means that
2
1 2 1
FµνM3 ...Mm = Fµν , F = ε(J)Fµν F µν e−2σ(J) , (5.781)
m! 2
where the indices M3 , . . . Mns belong to Js . Fµν F µν is written here in terms of the gµν , the 4-
dimensional part of gM N . If, as in Sec. 5.7.5, the observable metric is assumed to be hµν = e2f gµν ,
then the squared observed radial electric field strength is
1
E 2 [f ] = −htt huu (Fut )2 = − e−4f g tt g uu (Fut )2 = − e−4f Fµν F µν (5.782)
2
since Fut = −Ftu is the only nonzero component of Fµν . From (5.782) with (5.781) and (5.780) we
obtain
0
E 2 [f ] = Q2 e−4f −4β · e−4λϕ+2σ(J)−2σ(J) (5.783)
0
where e2β = gθθ , the notations (5.728) are used and J = {2, . . . , n} \ J.
CBPF-MO-003/02 123
0
One can notice that e2f +2β = hθθ = r2 where r is the observable radius of coordinate spheres
t = const, u = const. Therefore Eq. (5.783) may be rewritten as
This is the modified Coulomb law. The deviations from the conventional Coulomb law are
evidently both due to extra dimensions (depending on the multidimensional structure of the F -
form) and due to the interaction with the scalar fields. This relation (generalizing the one obtained
in Ref. [100] in the framework of dilaton gravity) is valid for an arbitrary metric of the form (5.210)
(d0 = 2) and does not depend on whether or not this F -form takes part in the formation of the
gravitational field.
Eq. (5.784) is exact and — which is remarkable — it is CF-independent. This is an evident
manifestation of the conformal invariance of the electromagnetic field in M phys even in the present
generalized framework.
From the observational viewpoint, the weak gravitational field approximation, in the spirit of
the previous section, can be of interest. Consider for simplicity the conventional Maxwell field,
i.e., the 2-form FM N with the only nonzero component Fut = −Ftu , so that J = ∅, σ(J) = 0 and
σ(J) = σ2 . Under the assumptions of Sec. 5.7.5, for large radii r (small u)
5.7.7 T-holes
Consider a simple example of a BH metric, e.g., (5.757) in the case d0 = 2, Q = 0, hence p = 0 and
H ≡ 1. It is thus a direct generalization of the Schwarzschild metric, charged p-branes are absent
and all extra dimensions form a 7-dimensional manifold with the r-independent Ricci-flat metric
ds27 , which we will assume to be flat. Let us introduce the following modification: interchange the
time t with a selected extra coordinate, say, v. One has
dr2 ³ 2k ´
ds211 = −dt2 + + r2 dΩ2 + 1 − ηv dv 2 + ds26 (5.786)
1 − 2k/r r
where ds26 is the flat metric of the remaining dimensions and ηv = ±1 depending on whether the
coordinate v is spacelike or timelike.
This modification only changes the interpretation of different coordinates without changing the
mathematical properties of the metric, it therefore remains to be a solution of the field equations.
The main feature of this configuration is that the physical space-time M phys changes its signature
at r = 2k: it is (+ − −−) for r > 2k and (+ + −−) at R < 2k. This evidently means that
the anomalous domains should be characterized with quite unconventional physics whose possible
consequences and observational manifestations are yet to be studied. It has been suggested [362] to
call such domains with an unusual space-time signature time holes or T-holes and the corresponding
horizons T-horizons, to be designated HT .
CBPF-MO-003/02 124
Evidently each BH configuration of any dimension D > 4 has a family of T-hole counterparts (a
family since the factor spaces may have different dimensions and signatures, and a v-axis like the one
in (5.786) may be selected in any of them). Conversely, any T-hole solution has BH counterparts.
If a BH possesses an external field, such as the F[4] -form field corresponding to the metric (5.757),
under a BH—T-hole transition, its true electric or magnetic component may be converted into a
quasiscalar one. This happens if the new t coordinate (former v) is off-brane, Rt 6∈ M I . If the new
time axis belongs to M I , the p-brane remains true electric or magnetic.
Unlike a BH-horizon, a T-horizon HT is not in absolute past or future from a distant observer’s
viewpoint, it is visible since it takes a finite time for a light signal to come from it (independently
of a conformal gauge since the latter does not affect light propagation).
Thus, in addition to the above family of p-brane BH solutions, there is a similar family of T-hole
ones.
There are certain problems connected with the compactification of extra dimensions. They can
be clearly understood using the simple example (5.786), which may be called the T-Schwarzschild
metric. Note that if we ignore the “passive” subspace with the metric ds26 , the remaining 5-
dimensional manifold coincides with the “zero dipole moment soliton” in the terminology of [95].
At r = 2k the signs of grr and gvv change simultaneously. Moreover, if ηv = −1, i.e., this
compactified direction is timelike at large r, the total signature of M is preserved but in the
opposite case, ηv = −1, it changes by four: two spacelike directions become timelike. However, as
one can directly verify, HT is not a curvature singularity, either for the D-dimensional metric or for
its 4-dimensional section.
If ηv = −1, the surface r = 2k is a Schwarzschild-like horizon in the (r, v) subspace, and there
exists an analytic continuation to R < 2k with the corresponding Kruskal picture. However, if some
points on the v axis are identified, as should be done to compactify the axis Rv in the conventional
way, then the corresponding sectors are cut out in the Kruskal picture, so that the T -domain and
R-domain sectors join each other only at a single point, the horizon intersection point. This should
be probably interpreted as a singularity due to intersection of particle trajectories.
Another thing happens if ηv = 1. Again a further study is possible using a transition to
coordinates in which the metric is manifestly nonsingular at r = 2k. Let us perform it for (5.786)
in the vicinity of HT (more general T-holes may be treated in a similar way):
r → 2k; r − 2k = (x2 + y 2 )/(8k); v = 4k arctan(y/x);
r − 2k 2 2k
ds22 (r, v) ≈ dv + dr2 = dx2 + dy 2 . (5.787)
2k r − 2k
Thus the (r, v) surface metric is locally flat near the T-horizon r = 2k which is transformed into
the origin x = y = 0, while the v coordinate has the character of an angle.
This transformation could also be conducted as a conformal mapping of the complex plane with
the aid of the analytic function ln z, z = x + iy, as was done in Ref. [374] for some cylindrically
symmetric Einstein-Maxwell solutions; then v is proportional to arg z.
Consequently, in the general case the (r, v) surface near r = 2k behaves like a Riemann surface
having a finite or infinite (if v varies in an infinite range) number of sheets, with a branch point at
x = y = 0 (a branch-point singularity [374]). If Rv is compactified, v is naturally described as an
angular coordinate (0 ≤ v < 2πl, where v = 0 and v = 2πl are identified and l is the compactification
radius at the asymptotic R → ∞). r = 2k is then the center of symmetry in the (r, v) surface; the
surface itself has the shape of a tube with a constant thickness at r → ∞, becoming narrower at
smaller r and ending at r = 2k either smoothly (if the regular center condition l = 4k is satisfied),
or with a conical or branch-point singularity (otherwise). This suggests that there is no way to go
beyond r = 2k.
CBPF-MO-003/02 125
bodies with mirror surfaces, at least if the T-horizons are connected with compact spacelike extra
dimensions. It is also possible that matter simply escapes from our physical space across the T-hole
surface. More detailed predictions can be formulated in specific theories. Although it seems hard to
point out a T-hole formation mechanism which might act in the present Universe, their emergence
should have been as probable as that of black holes in the early Universe, when all space-time
dimensions were on equal footing.
CBPF-MO-003/02 127
The model
We first consider a model governed by the action
Z p Z p
D
S = d z |g|{R[g] − 2Λ} + dD z |gB |σ, (5.788)
M D = R × M d, (5.789)
We start with the simplest solutions to the equations of motion corresponding to the action
(5.280) defined on the manifold (5.281) with Λ and σ of the form (5.283). The solution reads
B MN
T = TM Ng . Using the relation for the tt-component of the Ricci tensor of the metric (5.284)
1
with g from (5.290) (see Appendix B)
Rtt [g] = e2(φ+v) [e−2φ ∆v + φ00 + d(φ0 )2 + v 00 + (d + 1)v 0 φ0 + O(v 2 )], (5.800)
where d¯ = d − 2 and
ρ00 + (d + 1)φ0 ρ0 + m2 e−2φ ρ = 0. (5.803)
G is the (d-dimensional) gravitational constant, M is the mass and m is a spectral parameter. We re-
strict ourselves to the ”Yukawa-type” part of the general solution. We thus consider a superposition
of the Newtonian potential and the generalized Yukawa-type one.
Multitemporal generalization
Consider a multitemporal generalization of the metric (5.290), i.e.
n
X
1
g =− exp(2vi (~x, y))dti ⊗ dti + (d~x ⊗ d~x)d−n , (5.804)
i=1
i = 1, . . . , n. Here ∆ is the Laplace operator on Rd−n . For n = 1 we get Eq. (5.293). Eqs. (5.298)
can be easily solved in the symmetric case, when vi = v. In this case we get the relations (5.294)
with d¯ = d − 1 − n and (5.295) with d replaced by d + n. We note that in the vacuum case φ = 0
the multitemporal analogs of the Schwarzschild and Tangherlini solutions ¿from [71] give us exact
solutions to the field equations.
CBPF-MO-003/02 130
M 1 M
RN − δN R = κ2 TNM , (5.809)
2
where κ2 is the gravitational constant and the energy-momentum tensor is adopted in the form
Here py = py (y) is the pressure in the 1-dimensional y-space R and pi = pi (y) is the pressure in Mi ,
i = 1, . . . , n.
Using relations for Einstein tensor components EN M
= RNM
− 12 δN
M
R for the metric (5.299) (see
Appendix C), we obtain from Einstein Eqs. (5.301)
1
py = − 2
Gij φi,y φj,y , (5.811)
2κ
1
pi = −py − Gij (φj,yy + φj,y dk φk,y ), (5.812)
κ2 di
i = 1, . . . , n. Here
Gij = di δij − di dj , (5.813)
are components of the minisuperspace metric, i, j = 1, . . . , n [55]. The minisupermetric has a
pseudo-Euclidean signature.
The pressures may be decomposed into ”bulk” and ”brane” parts
pi = pbulk
i + pbr
i , (5.814)
where by definition
1
pbulk
i = −py − Gij φj,y dk φk,y , (5.815)
κ2 di
1
pbr
i = − Gij φj,yy , (5.816)
κ2 di
i = 1, . . . , n. Defining pbulk
y = py , pbry = 0, we decompose the stress-energy tensor into a sum of two
M bulk,M br,M
components: TN = TN + TN , where
bulk,M bulk m1 mn br,M m1 br mn
(TN ) = diag(py , p1 δk1 , . . . , pbulk
n δkn ), (TN ) = diag(0, pbr
1 δk1 , . . . , pn δkn ). As we shall see
below these decompositions will be justified by the thin brane limit.
CBPF-MO-003/02 131
As we shall see below, this condition is satisfied for isotropic case when all pressures are equal:
pi = p, i = 1, . . . , n.
For the pressures (5.307) and (5.308) we obtain in the special case (5.309)
µ ¶
bulk < uΛ , u > ui
pi = − 1 py , (f 0 (y))2 = 2py (y) (5.821)
< u, u > di
ui
pbr
i = √ f 00 . (5.822)
κdi − < u, u >
Here the vector
uΛi = 2di , (5.823)
i = 1, . . . , n, corresponds to the cosmological constant case.
Since the relations for the metric and pressures are invariant under the replacement u 7→ λu
(λ > 0) we may normalize u = (ui ) by the condition
and √
D − 2 00
pbr
i = √ f , pbulk
i = py = (f 0 (y))2 /2, (5.828)
κ D−1
i = 1, . . . , n. Since all scale factors are equal, one can reduce the isotropic case to the 1-space case
making the redefinition g 1 +. . . g n = ḡ 1 . The varying cosmological constant reads: Λ(y) = −κ2 py (y).
For n = 1 we get the relations (5.285) with 2pbr 1 = σ.
Thin brane. Consider a special solution with
p
φi (y) = 2py |y|, (5.829)
i = 1, . . . , n. We see that the thin-brane tension is proportional to a square route of the pressure in
the y-dimension.
Newtonian approximation. Using a relation for Rtt [g] from Appendix B in the multispace
case with g 1 from (5.290), we get a modification of the Eq. (5.293) to the perfect-fluid case with
(n − 1) internal Ricci-flat spaces
n
X
2φ1 00 1 0
∆v + e {v + [(φ ) + di (φi )0 ]v 0 } = 0, (5.831)
i=1
where ∆ is the Laplace operator on Rd1 −1 . Here all information about the perfect fluid and the
internal spaces is hidden in the behavior of functions φi (u). The formal solution (5.323) coincides
with that of (5.294) (d = d1 ) but the equation for the density (5.295) should be modified as follows
n
X 1
ρ00 + [(φ1 )0 + di (φi )0 ]ρ0 + m2 e−2φ ρ = 0. (5.832)
i=1
Appendix
A) Energy-momentum tensor for thick brane
Consider the thick brane part of the action
Z p
SB [g] = dD z |gB |σ(z) (5.833)
B) Ricci-tensor components
(1 + d)-dimensional case. Nonzero Ricci tensor components for the metric (5.284) are
Rtt [g] = e2(φ+v) {e−2φ [∆v + (5v)2 ] + φ00 + v 00 + (φ0 + v 0 )(dφ0 + v 0 )}. (5.841)
Multitemporal case. For the metric (5.284) with g 1 from (5.296) we obtain
n
X n
X
2(φ+vi ) −2φ 00
Rti ti [g] = e {e [∆vi + 5vi 5vj ]+φ +vi00 +(d−n)φ0 (φ0 +vi0 )+(φ0 +vi0 ) (φ0 +vj0 )}, (5.842)
j=1 j=1
CBPF-MO-003/02 134
C) Einstein tensor
Let us present expressions for the Einstein tensor EM N = RM N − 21 gM N R corresponding to the
metric (5.299) (see Appendix in [87])
i = 1, . . . , n, where
1
L= exp(γ0 )Gij (φi )0 (φj )0 (5.846)
2
Pn
is the Lagrangian and γ0 = i=1 di φi .
CBPF-MO-003/02 135
defined on RK , where here x = (xA ) = (−σ) ∈ RN , Φ = (Φs , s ∈ S) ∈ R|S| . The space M = (RK , G)
is a target space for the σ-model from the Sect. 2.
Here we prove that M is a homogeneous space isomorphic to G/H [213], where G is the isometry
group of M, and H is an isotropy subgroup of G. We prove that M is a symmetric space, iff (if
and only if) any two vectors U s1 and U s2 , s1 , s2 ∈ S, s1 6= s2 , are either coinciding U s1 = U s2 or
orthogonal (U s1 , U s2 ) = 0, with respect to the scalar product (3.82).
The Killing equations are solved in Subsect. 6.3. In Subect. 6.4 the block-orthogonal decompo-
sition of the set of U s -vectors is considered and under rather general assumptions the decomposition
of the Killing algebra into a sum of mutually commuting subalgebras is obtained.
Φs 7→ Φs + ∆s , xA 7→ xA , (6.848)
Φs 7→ Φs exp(−UAs δ A ), xA 7→ xA + δ A (6.849)
(U s1 , U s2 )(U s1 − U s2 ) = 0, (6.850)
for all s1 , s2 ∈ S.
Proof. Relation 5M RM1 M2 M3 M4 = 0 is equivalent to the relation (??) from Appendix 5, that is
equivalent to (??).
CBPF-MO-003/02 136
5M vN + 5N vM = 0, (6.851)
A, B = 1, . . . , N ; s, s0 ∈ S.
We note that the vector fields defined by relations (??)-(??) satisfy the Killing equations for
arbitrary U s -vectors. The sketch of proof of the Proposition 4 is given in Appendix 5. Here and in
what follows we assume that the restrictions (??) and (??) are satisfied.
The Killing vector fields corresponding to isometries (??) and (6.849) respectively are the fol-
lowing
Ts = ∂ s , (6.857)
X
DA = ∂A − UAs Φs ∂s , (6.858)
s∈S
s ∈ S; a = 1, . . . , p; α = p + 1, . . . , N .
The generators of generalized (pseudo) rotations are
Mαβ = xα ∂β − xβ ∂α , (6.862)
xα = ηαα xα , α, β = p + 1, . . . , N .
CBPF-MO-003/02 137
h0 =< Mαβ > is the isotropy subalgebra of g0 and hr =< Fr −θr Da(r) >, where θr ≡ εr sign(U r , U r ),
r ∈ ∆. The subalgebra hr is embedded into sl(2, R) as so(2, R) for θr = +1, and as as so(1, 1, R),
for θr = −1 (this follows from the matrix realization of e+ , e− , h).
CBPF-MO-003/02 138
Under assumptions mentioned above the target space manifold may be decomposed as follows
Y Y
RN = RN −p × Mr × Gl , (6.874)
r∈∆ l∈∆+
M 1 M
RN − δN R = κ2 TNM (7.881)
2
M (m)
(κ2 is gravitational constant) imply 5M TNM = 0 and consequently 5M TN = 0.
We suppose that for any α-th component of matter the pressures in all spaces are proportional
to the density
(α) (α)
pi (t) = (1 − hi (x(t)))ρ(α) (t), (7.882)
where
(α) 1 ∂ (α)
hi (x) = Φ (x), (7.883)
di ∂xi
i = 1, . . . , n, where Φ(α) (x) is a smooth function on Rn , α = 1, . . . , m.
Xn
i i i i
Rmi ni = gm i ni
[ξi + exp(2x − 2γ)(ẍ + ẋ ( Ni ẋi − γ̇))], (7.885)
i=1
CBPF-MO-003/02 140
i = 1, . . . , n.
We put
n
X
γ = γ0 ≡ d i xi (7.886)
i=1
in (7.876) (the harmonic time is used). Then it follows ¿from (7.884) and (7.885) that the Einstein
equations (7.881) for the metric (7.876) with γ from (7.886) and the energy-momentum tensor from
(7.878), (7.879) are equivalent to the following set of equations
Xm
1
Gij ẋi ẋj + Vc + κ2 ρ(α) exp(2γ0 ) = 0, (7.887)
2 α=1
m
X n
X
i i 2 (α) i
−1 (α) (α)
ξi + ẍ exp(2x − 2γ0 ) = κ exp(2x ) [pi + (D − 2) (ρ − dj pj )], (7.888)
α=1 j=1
We impose the conditions of state in the form (7.882), (7.883). Then eq. (7.890) gives
where A(α) = const and eqs. (7.887), (7.888) may be written in the following manner
Xm
1
Gij ẋi ẋj + V + κ2 A(α) exp Φ(α) = 0, (7.892)
2 α=1
m
X
i i 2
ξi + ẍ exp(2x − 2γ0 ) = −κ ui(α) A(α) exp(2xi − 2γ0 + Φ(α) ), (7.893)
α=1
i = 1, . . . , n. In (7.893) we denote
(α) (α) (α)
ui ≡ di hi = ∂i Φ(α) , ui(α) = Gij uj , (7.894)
z a = eai xi , (7.907)
´
(We note that in notations of [54] u2 = ∆(h)/(2 − D).)
This proposition follows from the fact that
is bilinear symmetric 2-form of signature (−, +, . . . , +) and the following quite obvious
Proposition 7.1.1a. Let v ∈ E = Rn , n ≥ 2, and < ., . >: E × E −→ R is a bilinear
symmetric 2-form of signature (−, +, . . . , +). Then there exists a basis e0 , . . . , en−1 in E, such that
< ea , eb >= η ab and a) e = e0 , b) e = e1 , c) e = e0 + e1 , in the cases: a) e2 ≡< e, e >= −1, b)
e2 = 1, c) e2 = 0, respectively.
Let u 6= 0. In z = (z a )-coordinates (7.907) with the matrix (eai ) from the Proposition 7.1.1 the
Lagrangian (7.895) has the following form
X1 n−1
1 1
LA = ηab ż a ż b − VA = − (ż 0 )2 + (ż i )2 − VA , (7.912)
2 2 i=1
2
CBPF-MO-003/02 143
where
z̈ a = −η ab ∂b VA (7.917)
z i = pi t + q i , i = 1, . . . , n − 1, (7.919)
2qz 0 = y(t), (7.920)
z i = pi t + q i , i = 0, 2, . . . , n − 1, (7.924)
2qz 1 = y(t), (7.925)
P
with (~p)2 = (p0 )2 − n−1 i 2
i=2 (p ) in (7.920)-(7.923).
c) u2 = 0, u 6= 0. In this case
z i = pi t + q i , i = 2, . . . , n − 1, (7.926)
z = z + z = p t + q+,
+ 0 1 +
(7.927)
z − = z 0 − z 1 = p− t + q − + κ2 Az(t), (7.928)
where for p+ 6= 0
z(t) = 2(p+ )−2 exp(p+ t + q + ), p+ p− = (~p)2 (7.929)
(p− = 0 for n = 2) and for p+ = 0
P
Here (~p)2 = n−1 i 2
i=2 (p ) .
For u = 0 we have
z a = pa t + q a , a = 0, . . . , n − 1, (7.931)
1
η p p + κ2 A = 0.
2 ab
a b
(7.932)
Kasner-like parametrization. Here we consider the case u2 < 0, A 6= 0. For C = −u2 (~p)2 > 0
we reparametrize the time variable
√ √
T exp( 12 C(t − t0 )) + ε
τ = √ ln √ √ , (7.933)
ε exp( 12 C(t − t0 )) − ε
where
ε ≡ A/|A| = ±1, T ≡ (2/κ2 |A||u2 |)1/2 . (7.934)
We introduce new (Kasner-like) parameters
p
β i ≡ −2eis ps / −u2 (~p)2 , (7.935)
where (eia ) = (eai )−1 and the summation parameter s runs: s = 1, . . . , n − 1. Then, due to relations
(7.907), (7.909), (7.919)-(7.921), (7.923) and Proposition 7.1.1 we get the following expression for
the metric (7.876)
Yn n
X
g = −( (ai (τ ))2di −ui )dτ ⊗ dτ + a2i (τ )ĝ i , (7.936)
i=1 i=1
(see Proposition 7.1.1 and (7.935)). For the density (7.896) we have
n
Y
ρ(τ ) = A (ai (τ ))ui −2di . (7.940)
i=1
Q
We note, that (~p)2 = 2κ2 |A| ni=1 (Ai )ui .
For A > 0 we have an exceptional solution (7.936), (7.938), (7.939) with the scale factors
Āi > 0, i = 1, . . . , n. This solution correspond to C = 0 case (7.922). Here we use the following
time parametrization
T /(t − t0 ) = exp[±τ /T ], (7.942)
CBPF-MO-003/02 145
t > t0 .
7.1.2.2. Two spaces with m-component matter.
Here we consider the following case: n = 2, m ≥ 2, A(α) 6= 0,
where 2α+ = − < u(1) , u∗ >, 2α− = − < u(1) , u >, and u∗ = (u∗i ) is defined by the relation :
u∗i xi = z − (or equivalently < u∗ , u∗ >= 0, < u∗ , u >= −2).
Due to (7.943)-(7.945) the potential in (7.895) is factorized
V = V+ (z + )V− (z − ), (7.946)
where
m
X
V+ (z + ) = exp(α+ z + )(κ2 A(1) + κ2 A(α) exp(bα z + )), (7.947)
k=2
− −
V− (z ) = exp(α− z ). (7.948)
F = e2f = V. (7.949)
w± (t) = p± t + q ± . (7.953)
In z-coordinates (7.907), corresponding to the case a) from the Proposition 7.1.1, the Lagrangian
(7.895) has the form (7.912) with the potential
m
X
0
VA = VA (z ) = κ2 A(α) exp(2qbα z 0 ), (7.955)
i=1
where q is defined in (7.916) (A = (A(α) )). The solutions of the equations (7.917) and (7.918) are
expressed by the formula (7.919) and the following relation
Z z0
dx[2E + 2VA (x)]−1/2 = ±(t − t0 ), (7.956)
c0
Pn−1
where 2E = i=1 (pi )2 , and c0 , t0 are constants.
where Ψ = Ψ(x) is ”the wave function of the Universe”, V is the potential (7.896), ∂i = ∂/∂xi . The
relation (7.957) is a result of a trivial quantization of the zero energy constraint (7.897), written in
the form µE = 0. Here µ is a fundamental quantum parameter of the theory.
In f -gauge (7.903) the WDW equation should be written in the conformally covariant form
[55, 67] (such form of the WDW equation was discussed earlier by Misner [66])
µ ¶
1 an
− ∆[e G] + R[e G] + e µV Ψf = 0,
2f 2f −2f
(7.958)
2µ µ
where ∆[Ĝ] and R[Ĝ] are the Laplace-Beltrami operator and the scalar curvature of Ĝ respectively,
an = (n − 2)/8(n − 1) and
Ψf = exp[(2 − n)f /2]Ψ. (7.959)
Without loss of generality we put µ = 1 below.
7.1.3.1. One-component matter.
Here we find the quantum analogues of the classical solutions ¿from subsection 7.1.2.1, i.e. we
integrate the WDW equation µ ¶
1 ab ∂ ∂
− η + VA Ψ = 0. (7.960)
2 ∂z a ∂z b
with the potential (7.913)-(7.915). We note, that the WDW equation for 1-component model with
n Ricci-flat spaces was considered previously in [253].
a) u2 < 0. In this case the WDW equation (7.960) reads
" n−1
#
∂ ∂ X ∂ ∂
− + 2κ2 A exp(2qz 0 ) Ψ = 0. (7.961)
∂z 0 ∂z 0 i=1
∂z i ∂z i
CBPF-MO-003/02 147
where p~ = (p1 , . . . , pn−1 ) is a constant vector (generally from Cn−1 ), ~z = (z 1 , . . . , z n−1 ), p~~z ≡
P n−1 i
i=1 pi z . The substitution of (7.962) into (7.961) gives
" µ ¶2 #
∂
− 0
− 2κ2 A exp(2qz 0 ) Φ = 2EΦ, (7.963)
∂z
Pn−1
where 2E = i=1 p2i . Solving (7.963), we get two linearly independent solutions
√ 0
Φ(z 0 ) = Bν ( −2κ2 Aq −1 eqz ), (7.964)
√
where ν = i 2E/q = i|~p|/q, and Bν = Iν , Kν is modified Bessel function. We note, that
Y u /2 n
1
v = exp (qz ) = exp( ui xi ) =
0
ai i (7.965)
2 i=1
i
is a natural scale factor for the model (ai = ex ).
The general solution of eq. (7.961) has the following form
X Z
Ψ(z) = dn−1 p~ CB (~p)ΨpB
~ (z), (7.966)
B=I,K
where √ −1 qz 0
ΨB i~
p~
z
~ (z) = e Bi|~
p
2
p|/q ( −2κ Aq e ), (7.967)
and functions CB (B = I, K) belong to an appropriate class of (generalized) functions.
b) u2 > 0. In this case the WDW equation (7.960) reads
" n−1
#
∂ ∂ ∂ ∂ X ∂ ∂
− 1 1+ 0 0− i ∂z i
+ 2κ2 A exp(2qz 1 ) Ψ = 0. (7.968)
∂z ∂z ∂z ∂z i=2
∂z
An analogous consideration in this case gives the general solution (7.966) with
√ −1 qz 0
ΨB ~ (z) = e Biν(~
p
i~
p~
z 2
p) ( 2κ Aq e ). (7.969)
√ Pn−1 2
Here p~ = (p0 , p2 , . . . , pn−1 ), ~z = (z 0 , z 2 , . . . , z n−1 ), ν(~p) = i 2E/q, and 2E = p20 − i=2 pi .
2
c) u = 0 for u 6= 0 the WDW equation reads
" n−1 µ ¶2 #
X ∂ 2 +
−4∂+ ∂− + + − 2κ A exp(z ) Ψ = 0, (7.970)
i=1
∂z i
Pn−1
where 2E = i=2 p2i . Introducing new variables u0 , u1 , where u0 ± u1 = u± and
u+ = 2Ez + + 2κ2 A exp(z + ), u− = z − , (7.973)
we get the Klein-Gordon equation for Φ with m2 = 1
"µ ¶2 #
∂ ∂ 2
− ( 1 ) + 1 Φ = 0. (7.974)
∂u0 ∂u
where
Ψp∗~ (z) = exp(i~p~z)Φp∗~ (z 0 ), (7.982)
∗ = ±, and Φ∗ = Φp∗~ (z 0 ) are two linerly independent solutions of the equation
" µ ¶2 #
∂
− − 2VA (z 0 ) Φ = 2Ep~ Φ, (7.983)
∂z 0
with the notations for E = Ep~ , p~, ~z from 7.3.1.a).
We note, that for special values of parameters A(α) and bα in the potential (7.955) the equation
(7.983) describes the quantum spin systems [250].
where ui = const, i = 1, . . . , n.
We impose also the following restriction on the vector u = (ui ) ∈ Rn
i = 1, . . . , n, where metric and ρ(τ ) are defined by (7.997) and (7.1000), respectively. Here Āi > 0
are constants, and T is defined in (7.1001).
For A > 0 the solution (7.1003) with the sign ” + ” is an attractor for the solutions (7.998), i.e.
where
ν i = 2ui / < u(Λ) − u, u >∗ (7.1011)
and for the density
−2 < u, u >∗
κ2 ρ = κ2 ρ(ts ) = Aρ t−2
s , Aρ = . (7.1012)
< u(Λ) − u, u >2∗
These solutions are called as the ”power-law inflationary” solutions. Here we use the relation
n
Y
Aiui −2di = T 2 /(σ̄ − 1)2 = Aρ /A (7.1013)
i=1
CBPF-MO-003/02 152
(see 7.1001)).
For
σ̄ = 1 ⇐⇒ < u(Λ) − u, u >∗ = 0 (7.1014)
we get ρ = const and · ¸
ui ts
ai (ts ) = Āi exp ∓ √ , (7.1015)
− < u, u >∗ T0
where T0 = (2κ2 ρ)−1/2 . Here we are led to so-called ”exponential inflation”.
Kasner-like behaviour in non-exceptional case. It was shown in [103] that ts → +0 (up
to a constant shift) as τ → +0 for
Xn
A) ui > 0, (7.1016)
i=1
and ts → +∞ as τ → +0 for
n
X
B) ui < 0. (7.1017)
i=1
as ts → +0 in the case A) and as ts → +∞ in the case B). Here B̄i , B̄ϕ are constants and parameters
αi , αϕ satisfy the Kasner-like relations
n
X n
X
i
di α = 1, di (αi )2 + αϕ2 = 1. (7.1019)
i=1 i=1
as ts → +∞, where ν i are defined in (7.1011). In (7.1022) and (7.1023) the parameter Aρ from
(7.1012) is used. For σ̄ < 0 we have the asymptotic behaviour (7.1023) in the limit ts → +0. In all
cases scalar field tends to constant: ϕ(ts ) ∼ const.
Isotropic case. Let us consider the isotropic case
ui = hdi (7.1024)
and hence
σ i = 2/[h(D − 1)] = σ(h). (7.1026)
The solution (7.997)-(7.1002) reads
Q P
g = −( ni=1 (ai (τ ))(2−h)di )dτ ⊗ dτ + ni=1 a2i (τ )ĝ i , (7.1027)
βi
ai (τ ) = Ai [sinh(rτ /T )/r]σ(h) [tanh(rτ /2T )/r] , (7.1028)
exp(κϕ(τ )) = Aϕ [tanh(rτ /2T )/r]βϕ , (7.1029)
Q (h−2)di
κ2 ρ(τ ) = A( ni=1 Ai )[sinh(rτ /T )/r]2(h−2)/h , (7.1030)
p −1/2
i = 1, . . . , n; where r = A/|A|, T = |h|−1 {κ2 |A|(D − 1)/[2(D − 2)]} , Ai , Aϕ > 0 are constants
i
and the parameters β , βϕ satisfy the relations
n
X n
X
i 4(D − 2)
di β = 0, di (β i )2 + (βϕ )2 = . (7.1031)
i=1 i=1
h2 (D − 1)
Example: Λ-term case [112]. For the cosmological term case corresponding to h = 2 we
obtain from (7.1027)-(7.1029)
P
g = −dts ⊗ dts + ni=1 a2i (ts )ĝ i , (7.1032)
i
ai (ts ) = Ai [sinh(rts /T )/r]1/(D−1) [tanh(rts /2T )/r]β , (7.1033)
βϕ
exp(κϕ(ts )) = Aϕ [tanh(rts /2T )/r] , (7.1034)
p
i = 1, . . . , n; where r = Λ/|Λ|, T = [2|Λ|(D − 1)/(D − 2)]−1/2 , Ai , Aϕ > 0 are constants and the
parameters β i , βϕ satisfy the relations
n
X n
X (D − 2)
di β i = 0, di (β i )2 + (βϕ )2 = . (7.1035)
i=1 i=1
(D − 1)
Here κ2 ρ = Λ is constant.
Exceptional solutions. Now, we consider the exceptional solutions for A > 0. From (7.1008)
and (7.1025) we have
From (7.1036) we get: < u(Λ) − u, u >∗ = 0 ⇐⇒ h = 2 (here h 6= 0). The matter corresponds to
cosmological constant Λ = κ2 ρ > 0. The relations (7.1015), (7.1025) imply the solution [71, 112]
with metric (7.1007) and
p
ai (ts ) = Āi exp[±ts 2Λ/R], R = (D − 1)(D − 2). (7.1037)
Here ϕ = const.
For h 6= 2 ( ⇐⇒ < u(Λ) − u, u >∗ 6= 0)
From (7.1010), (7.1012), (7.1025), (7.1036) and (7.1038) we obtain the relations for scale factors
and density:
ai (ts ) = Ai tν(h)
s , κ2 ρ(ts ) = Aρ t−2
s , (7.1039)
where Aρ = 2(D − 2)/[(h − 2)2 (D − 1)]. For h < 2 (or p > −ρ) we have an isotropic expansion of
all scale factors and for h > 2 (or p < −ρ) we have an isotropicP contraction.
Kasner-like behaviour. In the case under consideration ni=1 ui = h(D − 1) and hence (see
(7.1016), (7.1017)) the Kasner-like behaviour (7.1018), (7.1019) takes place as: A) ts → +0 for
h > 0 (or p < ρ) and B) ts → +∞ for h < 0 (or p > ρ).
Isotropization-like behaviour. In this special case we are lead to the following attractor
behaviour:
ai (ts ) ∼ Ai tν(h)
s , κ2 ρ(ts ) ∼ Aρ t−2
s . (7.1040)
in the limits ts → +∞ for 0 < h < 2 (or −ρ < p < ρ) and ts → +0 for h < 0 (or p > ρ).
Example: dust matter [103]. For the dust matter case with h = 1 (p = 0) and ρ > 0, the
solution (7.1027)-(7.1031) has the following sinchronous-time representation
P
g = −dts ⊗ dts + ni=1 a2i (ts )ĝ i ,
i
ai (ts ) = Āi [ts (ts + T1 )]1/(D−1) [ts /(ts + T1 )]β /2 , (7.1041)
exp(2κϕ(ts )) = Aϕ [ts /(ts + T1 )]βϕ , (7.1042)
κ2 ρ(ts ) = 2(D − 2)/[(D − 1)ts (ts + T1 )]. (7.1043)
i = 1, . . . , n; where 0 < ts < +∞, T1 > 0, Āi , Aϕ > 0 are constants and the parameters β i , βϕ
satisfy the relations
X n Xn
4(D − 2)
di β i = 0, di (β i )2 + (βϕ )2 = . (7.1044)
i=1 i=1
(D − 1)
The special case of this solution with βϕ = 0 was considered previously in [54] (for n = 2 and
d1 = . . . dn = 1 see [41] and [42], correspondingly.)
Example: curvature-like component. Now we consider the perfect fluid matter with
(1)
ui = 2h(−δi1 + di ) = hui , (7.1045)
where h 6= 0 is constant and d1 > 1. For h = 1 this component corresponds to the curvature term
for the first space (see next chapter below). The calculation gives
2h i
ui = − δ, < u, u >∗ = −4h2 b1 < 0, (7.1046)
d1 1
CBPF-MO-003/02 155
1
where b1 = 1 − d1
and
δ1i
< u, u(Λ) >∗ = 2ui di = −4h, < u, u(Λ) − u >∗ = 4h(−1 + hb1 ), σ i = . (7.1047)
h(d1 − 1)
For h 6= h0 ≡ b−1
1 = d1 /(d1 −1) > 1 we have from (7.1047) < u, u
(Λ)
−u >∗ 6= 0 and (see (7.1011))
The scale factors of internal spaces in this solution are constant (we have the so-called ”spontaneous
compactification”). It is note difficult to show that constancy of internal scale factors leads to the
equation of state (7.1045).
Using the relations σ̄ = h0 /h, (7.1022) and (7.1023) we obtain that the solution (7.1054)-(7.1057)
is an attractor for non-exceptional solutions with ρ > 0 as ts → T∗ − 0, for h > h0 ; ts → +∞,
for 0 < h < h0 and ts → +0, for h < 0. So, we obtained the solutions with the ”dynamical
compactification”.
7.2.1.2. Solutions with pure imaginary scalar field.
Now, we consider the solutions of the field equations with the complex scalar field and the
real metric. Solutions with imaginary scalar field and negative energy density (e.g. instanton and
classical wormhole ones) occur different in models gravitational models, e.g. with axion field. The
interest to wormhole solutions was stimulated greatly after the papers [266, 267, 268]).
Let us consider pure imaginary β k
β k = iβ̂ k , (7.1058)
CBPF-MO-003/02 156
k = 1, . . . , n and real βϕ . The solution may be obtained from (7.997)-(7.1002) substituting (7.1058)
and τ /T 7→ τ /T + i π2 :
n
Y n
X
g = −( (ai (τ ))2di −ui )dτ ⊗ dτ + a2i (τ )ĝ i , (7.1059)
i=1 i=1
σi β̂ i
ai (τ ) = Âi [cosh(τ /T )] [f (τ /2T )] , (7.1060)
ϕ(τ ) = c + 2iβϕ arctan exp(−τ /T ), (7.1061)
The scalar field ϕ(t) varies from c + iπβϕ to c as τ varies from −∞ to +∞. The solution (7.1059)-
(7.1062) for τ ∈ (−∞, +∞) is non-singular . Any scale factor ai (τ ) for some τ0i has a minimum
and
ai (τ ) ∼ A± i
i exp(σ |τ |/T ), (7.1065)
for τ → ±∞.
The ”Lorentzian” solutions considered above have also ”Euclidean” analogues for A < 0
n
Y n
X
2di −ui
g=( (ai (τ )) )dτ ⊗ dτ + a2i (τ )ĝ i , (7.1066)
i=1 i=1
i i
ai (τ ) = Âi [cosh(τ /T )]σ [f (τ /2T )]β̂ , (7.1067)
ϕ(τ ) = c + 2iβϕ arctan exp(−τ /T ), (7.1068)
with the parameters β̂ i , βϕ satisfying the relations (7.1062). When all spaces (Mi , g i ) are Rieman-
nian, this solution may be interpreted as the classical Euclidean wormhole (instanton) solution.
An interesting special case of the solution (7.1066)-(7.1068) occurs for β̂ i = 0, i = 1, . . . , n,
i
ai (τ ) = Âi [cosh(τ /T )]σ , (7.1069)
ϕ(τ ) = c ± 2iq −1 arctan exp(−τ /T ). (7.1070)
All scale factors (7.1069) have a minimum at τ = 0 and are symmetric with respect to the time
inversion: τ 7→ −τ . It is necessary to stress that here, like in [112], wormhole solutions take place
only in the presence of the imaginary scalar field.
CBPF-MO-003/02 157
Here we put A > 0, i.e. the density of matter is positive. We consider the case of a real Ψ-field
as in [78] for simplicity. The WDW equation (7.1071) corresponds to the action
Z
1
S= dn+1 zΨĤΨ. (7.1072)
2
Let us consider two bases of the solutions of the WDW equation: {Ψin (~p), Ψ∗in (~p)} and {Ψout (~p), Ψ∗out (~p)}
· ¸1/2 Ã√ !
π 2A qz0
Ψin (~p) = Ψin (~p, z) = J−i|~p|/q e (2π)−n/2 exp(i~p~z), (7.1073)
2q sinh(π|~p|/q) q
µ ¶1/2 Ã√ !
1 π (2) 2A qz0
Ψout (~p) = Ψout (~p, z) = Hi|~p|/q e (2π)−n/2 exp(i~p~z), (7.1074)
2 q q
(2)
where Jν and Hν are the Bessel and Hankel functions, respectively. These solutions are normalized
by the following conditions
¡ ¢ ¡ ¢ ¡ ¢
Ψin (~p), Ψin (~p 0 ) = Ψout (~p), Ψout (~p 0 ) = δ p~ − p~ 0 , (7.1075)
where Z ³ ´
↔
(Ψ1 , Ψ2 ) = i dn~z Ψ∗1 ∂0 Ψ2 (7.1076)
↔
is the charge form (indefinite scalar product). Here Ψ1 ∂ Ψ2 = Ψ1 ∂Ψ2 − (∂Ψ1 ) Ψ2 . Due to
asymptotic behaviour
where Ψin (~p, z) and Ψout (~p, z) are negative frequency modes of ”Kasner”- and ”Milne”- types re-
spectively.
The standard quantization procedure [256, 257] give us
Z
£ ¤
Ψ(z) = dn p~ a+ p)Ψ∗in (~p, z) + ain (~p)Ψin (~p, z) ,
in (~
Z
£ ¤
= dn p~ a+ p)Ψ∗out (~p, z) + aout (~p)Ψout (~p, z) ,
out (~ (7.1079)
The modes (7.1073) and (7.1074) are related by the Bogoljubov transformation
The temperature (7.1085) depends upon the vector u = (ui ) ( i.e. on the equation of state):
TP l = TP l (u). For example, we get TP l (u(Λ) ) = 2TP l (u(dust) ). In the Zeldovich matter limit u → 0
we have TP l → +0.
Remark. Wick rotation [265]. In [260] a regularization of propagators (in quantum field
theory) was introduced using the complex signature matrix
where w ∈ C \ (−∞, 0] is the complex parameter (Wick parameter). Originally path integrals are
defined (in covariant manner) for w > 0 (i.e. in Euclidean-like region) and then should be analyti-
cally continued to negative w. The Minkowsky space limit corresponds to w = −1 + i0 (in notations
of [260] w−1 = −a). The prescription [260] is a natural realization of the Wick’s rotation. In [261]
the analogs of the Bogoljubov-Parasjuk theorems [255] for a wide class of propagators regularized
by the complex metric (7.1086) were proved. This formalism may be applied for third-quantized
models of the multidimensional cosmology. In this case the corresponding path integrals should be
analytically continued from the interval 1 <
PD < 2 (D is dimension), where minisuperspace metric
n
Gij is Euclidean, to D = D0 −i0, D0 = 1+ i=1 di . We note also that recently J.Greensite proposed
an idea of considering the space-time signature as a dynamical degree of freedom [262] (see also
[263, 264]).
M = R × M1 × . . . × Mn , (7.1087)
CBPF-MO-003/02 159
where γ(t) is an arbitrary function determining the time t. It is supposed that the manifold Mi (see
(7.212)) for i = 1, . . . , n is the Einstein space of dimension Ni with the metric g (i) , i.e.
Rmi ni [g (i) ] = λi gm
(i)
i ni
, mi , ni = 1, . . . , Ni , (7.1089)
where λi is constant. (For the manifold Mi of constant Riemann curvature Ki the constant λi reads:
λi = Ki (Ni − 1)).
The non-zero components of the Ricci tensor for the metric (7.213) are the following [61]
n
X
R00 = exp[−2γ] Ni [ẍi − γ̇ ẋi + (ẋi )2 ], (7.1090)
i=1
CBPF-MO-003/02 160
à " à n
!# !
X
Rklii = δklii λi exp[−2xi ] + ẍi + ẋi Nj ẋj − γ̇ exp[−2γ] , (7.1091)
j=1
Pi−1 P
where indices ki and li for i = 1, . . . , n run over from 1 + j=1 Nj to ij=1 Nj .
After the gauge fixing γ = F (xi ) vacuum Einstein equations RB A
= 0 for A, B = 1, . . . , D − 1
0 0
and equation R0 − δ0 R/2 = 0 may be presented as the Lagrange-Euler equations and zero-energy
constraint correspondingly obtained from some Lagrangian L(xi , ẋi ) = LF (xi , ẋi ). For the so-called
harmonic time gauge
Xn
γ = γ0 ≡ Ni x i (7.1092)
i=1
where
Gij = Ni δij − Ni Nj (7.1094)
are components of the minisuperspace metric. The minisuperspace metric has the pseudo-Euclidean
signature (−, +, . . . , +) [54, 55]. (This is a well-known property of the Hilbert-Einstein action [399].)
If all spaces M1 , . . . , Mn are Ricci-flat, i.e. λi = 0 for i = 1, . . . , n, or only one of them has non-
zero Ricci tensor, then equations of motion following from the Lagrangian (7.218) are integrable
and exact solutions are obtained in [61]. But nearly nothing is known about integrability of the
models for two or more Einstein spaces with non-zero Ricci tensor. Obtaining exact solutions for
such models is connected with a general problem of integrability of the so-called pseudo-Euclidean
Toda-like systems [61, 72]. These systems are described by the Lagrangian of the following form
n−1 m n−1
1 X a b
X
(s)
X
L= ηab ż ż − A exp[ usa z a ], (7.1095)
2 a,b=0 s=1 a=0
where all usa , A(s) are real constants and (ηab ) =diag(−1, +1, . . . , +1). The Lagrangian (7.218) may
be reduced to the Lagrangian (7.220) by some linear transformation of the configuration variables
x1 , . . . , xn . It should be noted, that the Lagrangian (7.220) describes both vacuum multidimensional
cosmological models or models with multicomponent perfect fluid source [72, 73]. The well-known
4-dimensional Bianchi-IX model (the so-called ”mixmaster model”) is also described by the La-
grangian (7.220) with n = 3, m = 6 and certain usa , A(s) [400].
There are extensive literature devoted to the integrability of the Euclidean Toda-like systems
((ηab ) =diag(+1, . . . , +1) in (7.220)) and methods of their explicit integration (see, for instance,
[226]), but, as far as we know, the integrability of the pseudo-Euclidean systems is not well studied
yet. In our previous paper [73] we singled out two integrable by quadratures classes of these
systems and developed their integration. The first class contains the pseudo-Euclidean systems
trivially reducible to the Euclidean systems. It is easy to see that under the conditions: us0 = 0
for s = 1, . . . , m it follows from (7.220) that ż 0 =const. Then the Lagrangian (7.220) effectively
”loses” negative term in the kinetic energy, so for z 1 , . . . , z n−1 the Euclidean system arises. The
second class appears provided vectors u1 , . . . , um ∈ Rn , where us = (us0 , . . . , usn−1 ), are collinear or
form the orthogonal set with respect to the symmetrical bilinear form on Rn with matrix (ηab ) (see
CBPF-MO-003/02 161
for details [73]). But the models describing evolution of two or more Einstein spaces with non-zero
Ricci tensors are out of these classes and until now we had neither any integrability conditions nor
explicit integration methods for them. In this paper we show that at least the models with n = 2, i.e.
multidimensional vacuum cosmological models with metric (7.213) on the manifold (7.212), where
M1 and M2 are Einstein spaces of certain dimensions with non-zero Ricci tensor, are integrable by
quadratures, and develop their integration procedure.
So, let us consider the Lagrangian (7.218) for n = 2 and λ1 , λ2 6= 0. After the following
coordinate transformation diagonalizing the minisuperspace metric:
¯ ¯
1 2 1 ¯¯ λ2 N2 ¯¯
x = x − x + ln ¯ , (7.1096)
2 λ 1 N1 ¯
¡ ¢
y = αβ N1 x1 + N2 x2 , (7.1097)
Here we denoted
1 N2 N1
V0 = − |λ1 N1 |β2 |λ2 N2 |−β1 , β1 = − , β2 = . (7.1101)
2 N1 + N2 N1 + N2
We note that β2 − β1 = 1. The corresponding zero-energy constraint reads
1 ¡ 2 2
¢
E≡ ẋ − ẏ + V (x, y) = 0. (7.1102)
2β 2
It is easy to see, that the set of equations (7.228), (7.229) does not admit solutions with ẋ = ẏ = 0
∀t. Indeed, such solutions are possible only if λ1 λ2 < 0 and β1 = β2 . But the latter condition is
impossible in the model under consideration (see (7.226)). So, our model has no static solutions for
any dimensions of M1 and M2 .
To integrate the equations of motion (7.228), (7.229) under the zero-energy constraint (7.227)
we consider at first the following procedure proposed in [60]. Using the relation
d2 y ÿ − ẍy 0
y 00 ≡ = , (7.1105)
dx2 ẋ2
CBPF-MO-003/02 162
equations of motion (7.228),(7.229) and the zero-energy constraint (7.227) in the form
V (x, y) V (x, y)
ẋ2 = 2β 2 2
= 2β 2 0 2 , (7.1106)
(ẏ/ẋ) − 1 (y ) − 1
where
After this procedure is done the following solutions may be lost: ẋ = ±ẏ ∀t and ẋ = 0 ∀t. It is
easy to see, that under the zero-energy constraint the equations of motion (7.228), (7.229) have no
solutions with ẋ = ±ẏ ∀t. The solutions with ẋ = 0 ∀t appear only in the case λ1 < 0 and λ2 < 0.
So, by integration of the equation (7.232) we obtain all possible trajectories on the configuration
plane (x, y) except the trajectory x = 12 ln[−β1 /β2 ] for the model with λ1 < 0 and λ2 < 0. If the
trajectories of motion are known we may get the law of motion for each trajectory by integration
of the equation (7.231). Thus, the problem of the integrability by quadratures of the equations
(7.228), (7.229) under the constraint (7.227) is reduced to the problem of the integrability of the
equation (7.232). In the paper [60] the equation (7.232) was studied by qualitative methods, but
here we consider its explicit integration.
For the considered model we have from (7.223) and (7.226)
N1 − N2 α R
β1 + β2 = , = , (7.1110)
N1 + N2 β N1 + N2
where p
R = R(N1 , N2 ) ≡ N1 N2 (N1 + N2 − 1). (7.1111)
We also rewrite (7.233), (7.234) as follows
exp(2x) − ε
f (x) = , (7.1112)
exp(2x) + ε
An equation of the form (7.239) is well-known as Abel’s equation (see, for instance [397, 398]).
There are no methods to integrate Abel’s equation with arbitrary functions f0 (x), f1 (x), f2 (x) and
f3 (x).
CBPF-MO-003/02 163
it may be written as
uu0 + F1 (x) + F2 (x)u = 0, (7.1116)
where
F1 (x) = f3 (x)(E(x))2 , F2 (x) = (3f3 (x)z0 + f2 (x)) E(x). (7.1117)
For the new variable Z
q=− F2 (x)dx (7.1118)
du
u − u = F (q), (7.1119)
dq
where we denoted
F1 (x)
F (q) = . (7.1120)
F2 (x)
The Abel equation in the form (7.244) was recently studied by a discrete-group analysis methods
and a number of new integrable equations were described [397, 398]. In principle, the Abel equation
(7.232) may be written in the canonical form as its solutions y 0 = ±1 are known, but some difficulty
to obtain the function F (q) in an explicit form for arbitrary parameters α, β and βi arises. It is not
hard to verify, that if one of the two following conditions holds
then the function F (q) for the equation (7.232) may be obtained in an explicit form by solving some
algebraic second order equation. Using (7.223) and (7.226 ) we get corresponding relations for the
dimensions of M1 and M2 .
4N2 4N1
N1 = , N2 = . (7.1122)
N2 − 1 N1 − 1
It can be easily seen that the first condition holds in the following three cases: i) N1 = 6 and N2 = 3;
ii) N1 = 8 and N2 = 2, iii) N1 = 5 and N2 = 5. The second condition corresponds to the inverse
numbering of the spaces M1 and M2 .
Under one of the conditions (7.247) the Abel equation (7.232) in the canonical form may be
reduced to the Emden-Fowler equation of the form: d2 Y /dX 2 =constY const [398], which may be
easily integrated.
(1+6+3)-model
First, let us consider only the model with N1 = 6 and N2 = 3. This (1 + 6 + 3)-model is of most
interest, because the 3-dimensional space M2 may be interpreted as our (external) space. Omitting
CBPF-MO-003/02 164
technical details we give at once the final Emden-Fowler equation for this model. One may check
that the Abel equation (7.232) for N1 = 6 and N2 = 3 by the transformation [398]
3X dY
1 + ε exp[2x] = , (7.1123)
2Y dX
3X dY
−2
y0 = 1 + Yh dX
¡ ¢ i, (7.1124)
X dY a dY −2
1+ Y dX Y dX
−1
By τ we denote the parameter (time), C1 and C2 are arbitrary constants. Formulas (7.251), (7.252)
together with y 0 = ±1 present the general solution to the Abel equation (7.232) for N1 = 6 and
N2 = 3. This may be also verified by a straightforward calculation using the following relation
δ £¡ 2
¢ ¤
Φ̇ = − 1 − 3δτ Φ + 1. (7.1131)
τ (τ 2 − δ)
The function (7.251) is depicted in Figs. 1, 2 for different values of δ and C1 . The limits Φ(τ ) → ∞
and Φ(τ ) → 0 (for non-exeptional C1 ) correspond to Kasner-like behaviour near the singularity (see
7.3.5 below).
Figs. 7.3.1, 7.3.2
Using relations (7.251), (7.256), zero-energy constraint (7.227) and the expression for the poten-
tial (7.225) we obtain the following relation between harmonic time and τ -variable
µ ¶2 µ ¶
dτ 2 −2 8
= A(−δε2 )|τ − δ| exp 4x + C2 , (7.1132)
dt 3
where
1
A = |6λ1 |2/3 |3λ2 |1/3 . (7.1133)
6
CBPF-MO-003/02 165
Using relations (7.265), (7.269) zero-energy constraint (7.227) and the expression for potential
(7.225) we obtain the following relation between harmonic time and τ -variable
µ ¶2 · µ ¶ ¸
dτ 2 1−k 1
= A(−δε2 )|τ − δ| exp 4x + 2 1 + C2 (7.1147)
dt k
where here
1 1 1
A= |N1 λ1 |1− k |N2 λ2 | k . (7.1148)
N1
From (7.272) we obtain the relation (7.259).
The transformation inverse to (7.221), (7.222) is the following
µ ¯ ¯¶
1 N2 1 ¯¯ λ2 N2 ¯¯ qy
x = x − ln ¯ ¯ + , (7.1149)
N1 + N2 2 λ1 N1 N 1 + N2
µ ¯ ¯¶
2 N1 1 ¯¯ λ2 N2 ¯¯ qy
x =− x − ln ¯ ¯ + , (7.1150)
N1 + N2 2 λ 1 N1 N1 + N2
where r
−1 N1 N2
q = (αβ) = . (7.1151)
N1 + N 2 − 1
Using (7.217), (7.259), (7.265), (7.267), (7.272)-(7.276) we get the relation for the metric (7.213)
(n = 2)
N1
g = c (f1 f2 )−r {− f1−2 dτ ⊗ dτ + f2 |λ1 |g (1) + |λ2 |g (2) }. (7.1152)
N2
Here c 6= 0,
1 1 2
r = q/N2 = 2 − q = , , (7.1153)
2 3 3
for (N1 , N2 ) = (6, 3), (5, 5), (8, 2) respectively , f1 = f1 (τ, ε2 ) is defined in (7.263) and
N1
f2 = f2 (τ, ε1 , ε2 , C1 , N1 , N2 ) = ε1 ε2 Φk (τ, ε2 , C1 ), (7.1154)
N2
where functions Φk are defined in (7.251), (7.270) and (7.271).
defined on the manifold (7.212), where (Mi , g (i) ) are Einstein spaces satisfying (7.214) (λi 6= 0) and
dimM = D, is a solution to the vacuum Einstein equations (or, equivalently, is Ricci-flat). (See
(7.215), (7.216)).
Remark 1. We note that the metric
n
X λi
g∗ = g (i) (7.1157)
i=1
(2 − D)
defined on the manifold M1 × . . . × Mn satisfies the same relation Rmn [g ∗ ] = −(D − 2)gmn
∗
as the
D−1
metric g(H ) of (D − 1)-dimensional Lobachevsky space. We also note that the metric
is flat. For D = 3 (7.283) coincides with the well-known Milne solution [401]. So, the metric (7.281)
may be called a ”quasi-flat” or ”Milne-like” metric.
Remark 2. It may be proved that, when (Mi , g (i) ) are spaces of constant (non-zero) curvature,
i = 1, ..., n, and n ≥ 2, the metric (7.281) has a divergent Riemann-tensor squared as ts → +0
In this case the solution (7.281) is singular. The relation (7.284) follows from the formula
I[g] = At−4
s , (7.1160)
where n
X I[g (i) ]
A = (D − 2)2 − 2(D − 1)(D − 2). (7.1161)
i=1
λ2i
(see [104, 84]). For spaces of constant negative curvature
à n !
X Ni D − 1
A = 2(D − 2)2 − > 0, (7.1162)
i=1
N i − 1 D − 2
if n ≥ 2.
Isotropization. Now, we show that the solution (7.280) is an attractor for solutions (7.277)
with ε1 < 0, ε2 < 0 and c > 0, as τ → 1 (or ts → +∞). Using the relation f2 (τ ) → 1 as τ → 1, we
get the asymptotical formula for the synchronous-time variable ts (c > 0)
p
1/2 N1 /N2
c ts ∼ |τ − 1|−r/2 as τ → 1. (7.1163)
r2r/2
Hence
r 2 N2 2 t2s
(f1 f2 )−r ∼ ts = as τ → 1. (7.1164)
N1 D−2
Thus, we obtain the relation for the metric
where
t2s
bi (ts ) ∼ as ts → +∞. (7.1166)
D−2
CBPF-MO-003/02 168
where Z µ ¶1/2
τ
0 N1
ts = ± dτ (f1 f2 )−r/2 f1−1 c + t0s , (7.1168)
τ0 N2
(fi = fi (τ 0 )). As follows from [87] almost all solutions for the considered model should have a
Kasner-like behaviour for small ts , i.e.
where Ai > 0 are constants and the Kasner parameters αi satisfy the relations
2
X 2
X
Ni α i = Ni αi2 = 1. (7.1170)
i=1 i=1
Here a suitable choice of the constant t0s = t0s (τ0 ) and the sign in (7.293), such that v = aN1 N2
1 a2 →
+0 corresponds to ts → +0, is assumed. We note, that although in [87] models with minisuperspace
dimensions n ≥ 3 were considered, the results are also applicable for the case n = 2.
Solving the equations (7.295) we get [112]
N1 ± R N2 ∓ R
α1 = , α2 = , (7.1171)
(N1 + N2 )N1 (N1 + N2 )N2
where R = R(N1 , N2 ) is defined in (7.236). For (N1 , N2 ) = (6, 3), (5, 5), (8, 2) R is integer
for (N1 , N2 ) = (6, 3), (5, 5), (8, 2) respectively. Here (α1 , α2 )± corresponds to ”±” in (7.296).
Now, we return to the metric (7.277) written in the synchronous-time form (7.292). The Kasner-
like behaviour (7.294) with αi from (7.298) takes place if τ → τ∗ , where f2 (τ∗ ) = 0 and f20 (τ∗ ) 6= 0.
Indeed, from (7.293) we get
r
ts ∼ const|τ − τ∗ |1− 2 as τ → τ∗ (7.1175)
(r is defined in (7.278)) and using (7.277) we obtain the Kasner-like behaviour (7.294) with the
parameters
1−r −r
α1 = , α2 = , (7.1176)
2−r 2−r
coinciding with those from eq. (7.298)
CBPF-MO-003/02 169
The Kasner-like asymptotes (7.294) with parameters ¿from (7.299) take place, when τ → ±∞ ,
for non-exceptional values of the constant C1 :
π
C1 6= 0, if ε2 = −1 and C1 6= ∓ , if ε2 = +1, for (N1 , N2 ) = (6, 3), (8, 2); (7.1177)
2
C1 6= ±2 for (N1 , N2 ) = (5, 5). (7.1178)
In this case
f2 (τ ) ∼ B± |τ |k as τ → ±∞, (7.1179)
where B± are constants (integer k is defined in (7.268)) and
s
ts ∼ const|τ |−1− 2 as τ → ±∞. (7.1180)
Here s = (k + 2)r. From (7.277), (7.304) and (7.305) we get the Kasner-like behaviour (7.294) with
parameters
s−k s
α1 = , α2 = , (7.1181)
2+s 2+s
coinciding with (7.299).
According to the results of [104] the Riemann tensor squared for the solutions with asymptoti-
cally Kasner-like behaviour for ts → +0 is divergent
as ts → +0 for fixed (x1 , x2 ) ∈ M1 × M2 . If I[g (i) ] ≥ ci for some ci , i = 1, 2, then the relation (7.104)
takes place uniformly on (x1 , x2 ) ∈ M1 × M2 (see Theorem in [87]).
where µi 6= 0 is a real constant, i = 1, ..., k. Here and below we denote by Ric(h) the Ricci-tensor
corresponding to the metric h. Let µ 6= 0 be a real number. Then
k
X µi
h= h(i) (7.1184)
i=1
µ
M = M1 × . . . × Mk (7.1185)
and satisfying
Ric(h) = µh. (7.1186)
Indeed,
k
X X k X k
µi (i)
Ric(h) = Ric( h )= Ric(h(i) ) = µi h(i) = µh. (7.1187)
i=1
µ i=1 i=1
CBPF-MO-003/02 170
Remark 3. In (7.309) (like in (7.213)) we identify the metric h(i) on Mi with its canonical
extension to the manifold M (7.310). It is more correct to write (instead of (7.309))
k
X µi
h= p∗i h(i) , (7.1188)
i=1
µ
j = 1, . . . , ni ; i = 1, 2. Clearly that
ni
X
Ni = Nij (7.1192)
j=1
where
Nij = dimMij > 1, (7.1193)
j = 1, . . . ni ; i = 1, 2. It follows from (7.317), (7.318) that
· ¸
Ni
1 ≤ ni ≤ , (7.1194)
2
i = 1, 2; where [x] denotes the integer part of x.
Substituting (7.314) into (7.277) we get the following solutions to (vacuum) Einstein equations:
( n1 n2
)
N 1
X X
g = c(f1 f2 )−r − f1−2 dτ ⊗ dτ + f2 ε1 λ1j g (1j) + ε2 λ2l g (2l) , (7.1195)
N2 j=1 l=1
The solution (7.320), (7.321) describes the evolution of n = n1 + n2 Einstein spaces (Mij , g (ij) ),
satisfying (6.316)-(6.319). Here (N1 , N2 ) = (6, 3), (5, 5), (8, 2); c 6= 0, εi = ±1; r = r(N1 , N2 ) and
fi = fi (τ ) are defined in 7.3, i = 1, 2.
The relations (7.320), (7.321) give us
(1 + 5 + 3 + 2), (1 + 3 + 2 + 3 + 2) (7.1198)
CBPF-MO-003/02 171
|τ | > 1.
Now we introduce a new time variable defined by the relation
Z τ s
ρ 2
ln =− dτ 0 , (7.1204)
ρ0 τ0 5f12 f2
We put
ρ0 τ0 exp [−I(τ0 , +∞)] = 1. (7.1208)
Then ρ ∼ τ1 as τ → ∞ . The function ρ = ρ(τ ) (7.330) is smooth and monotonically decreasing on
(1, +∞) and may be analytically continued to {|τ | > T } ∪ {∞}, where
" ∞ µ ¶2k #
1 X 1
ρ(τ ) = 1+ ρk . (7.1209)
τ k=1
τ
For this special case we get from (7.262), (7.329), (7.336), (7.337)
© £ ¤ ª
g = cF1 (ρ) −F2 (ρ) dρ ⊗ dρ + ρ2 dΩ26 + |λ2 |g (2) , (7.1213)
where
holomorphic in {|w| < δ 2 } and smooth on (0, +∞). In terms of these functions the solution (7.338)
may written as ( )
7
X
g = cΦ1 (|~x|2 ) −Φ2 (|~x|2 ) dxi ⊗ dxi + |λ2 |g (2) , (7.1219)
i=1
P7 i 2
where |~x|2 = i=1 (x ) . Clearly, (7.344) is a smooth metric on the manifold
R 7 × M2 . (7.1220)
CBPF-MO-003/02 173
The solution (7.344), (7.345) is an extension of the special solution defined for τ ∈ (0, +∞) to
the semi-interval τ ∈ (0, +∞]. It may be interpreted in two ways. First, we may say that the
metric (7.344) describes an extension of the 4-dimensional R × M2 cosmological solution to the
case of 7-dimensional Euclidean time manifold. On the other hand it may be also considered as
a spherically-symmetric (O(7)-symmetric) solution with a curved time manifold M2 (of negative
curvature).
It seems likely that the procedure considered here may be also applied to the special solution
with C1 = π2 , ε1 = −1, ε2 = +1 for N1 = 6, N2 = 3 and for other special solutions in (1 + 8 + 2)-
and (1 + 5 + 5)-models. We may also obtain the ”Milne-like” solution, when (M1 , g (1) ) is the
6-dimensional Lobachevsky space (H 6 , g(H 6 )).
7.3.8 Discussion
Here we have integrated the vacuum Einstein equations for the model describing the evolution of two
Einstein spaces M1 and M2 with dimensions (N1 , N2 ) = (6, 3), (5, 5), (8, 2). To our knowledge these
are the first non-trivial cosmological solutions describing the evolution of more then one Einstein
spaces of non-zero curvatures.
The Kasner-like behaviour near the singularity (for ts → +0) is investigated. The Kasner
parameters (7.298), (7.299) are rational for all considered three cases. We may consider the following
hypothesis: the Abel equation (7.232) may be integrated by methods described in [397, 398] (or by
its extension) for the dimensions satisfying
p
R(N1 , N2 ) ≡ N1 N2 (N1 + N2 − 1) = m ∈ Z. (7.1221)
(m is integer). In this (and only in this case) the Kasner parameters (7.298), (7.299) are rational.
(The relation (7.346) is satisfied for (N1 , N2 ) = (13, 13), (25, 3), (25, 25), (41, 41) etc).
Here we also received some special solutions to Einstein equations, i.e. ”Milne-type” solutions for
arbitrary n (see (7.281)), and two-parametric families for 3 ≤ n ≤ 5 (see (7.320)). We also obtained
non-singular solutions with topology R7 × M2 (sec. 7.3.7). (It may be shown that the non-singular
solutions with topology RN1 +1 × M2N2 exist also for other dimensions (N1 , N2 ) = (5, 5), (8, 2).)
The considered solutions to the vacuum Einstein equations are defined on the manifold M
of dimension D =dimM = 10, 11. These solutions satisfy the equations of motion for D = 10
supergravity of superstring origin and for D = 11, N = 1 supergravity respectively. For certain
manifolds M1 , M2 it is possible to use the obtained here solutions for generation of other classical
solutions with non-zero matter fields (dilatonic, Kalb-Ramond etc) in D = 10, 11 supergravities.
(This may be done, for example, by using the duality transformations.)
CBPF-MO-003/02 174
M = R × M1 × . . . × Mn , (7.1222)
CBPF-MO-003/02 175
The product of one part of the manifolds gives the external 3-dimensional space and the remaining
part stands for so called internal spaces. The internal spaces are supposed to be compact. Further,
for sake of generality, we admit that dimensions Ni = dimMi for i = 1, . . . , n are arbitrary.
The manifold M is equipped with the metric
n
X
2γ(t)
g = −e dt ⊗ dt + exp[2xi (t)]g (i) , (7.1223)
i=1
where γ(t) is an arbitrary function determining the time t and g (i) is the metric on the manifold
Mi . We suppose that the manifolds M1 , . . . , Mn are the Einstein spaces, i.e.
(i)
Rki li [g (i) ] = λi gki li , ki , li = 1, . . . , Ni , i = 1, . . . , n, (7.1224)
where λi is constant. In the special case, when Mi is a space of constant Riemann curvature Ki the
constant λi reads: λi = Ki (Ni − 1) (here Ni > 1).
Using the assumptions (7.349) we obtain the following non-zero components of the Ricci tensor
for the metric (7.348) [61]
à n !
X
0 −2γ i 2
R0 = e Ni (ẋ ) + γ¨0 − γ̇ γ˙0 (7.1225)
i=1
© £ ¤ ª
Rnmii = λi exp[−2x ] + ẍ + ẋi (γ˙0 − γ̇) e−2γ δnmii
i i
(7.1226)
where we denoted n
X
γ0 = Ni x i . (7.1227)
i=1
P P
ni in (7.350),(7.351) for i = 1, . . . , n run over from (D − nj=i Nj ) to (D − nj=i Nj +
Indices mi andP
Ni ) (D = 1 + ni=1 Ni = dimM ).
We consider a source of gravitational field in the form of multicomponent perfect fluid. The
energy-momentum tensor of such source under the comoving observer condition reads
m̄
X M (µ)
TNM = TN , (7.1228)
µ=1
³ ´ ³ ´
M (µ) (µ)
TN = diag −ρ(µ) (t), p1 (t)δlk11 , . . . , p(µ)
n (t)δ kn
ln , (7.1229)
Furthermore we suppose that for any µ-th component of the perfect fluid the barotropic equation
of state holds ³ ´
(µ) (µ)
pi (t) = 1 − hi ρ(µ) (t), µ = 1, . . . , m̄, (7.1230)
(µ)
where hi = const. It should be noted that each µ-th component admits different barotropic
equations of state in the different spaces M1 , . . . , Mn . From the physical viewpoint this follows
from the separation of the internal spaces with respect to the external one and with respect to each
others.
M (µ)
One easily shows that the equation of motion 5M T0 = 0 for the µ-th component of the
perfect fluid described by the tensor (7.354) reads
n
X (µ)
ρ̇(µ) + Ni ẋi (ρ(µ) + pi ) = 0. (7.1231)
i=1
CBPF-MO-003/02 176
Using the equations of state (7.355), we obtain from (7.356) the following integrals of motion
" n
#
X (µ)
A(µ) = ρ(µ) exp 2γ0 − Ni hi xi = const. (7.1232)
i=1
M M
The Einstein equations RN − RδN /2 = κ2 TNM (κ2 is the gravitational constant), can be written
M
as RN = κ2 [TNM − T δN
M
/(D − 2)]. Further we employ the equations R00 − R/2 = κ2 T00 and Rnmii =
2 mi mi
κ [Tni − T δni /(D − 2)]. Using (7.350)-(7.355), we obtain for them
n
1X
Gij ẋi ẋj + V = 0, (7.1233)
2 i,j=1
m̄
à Pn !
X (µ)
−2xi (µ) k=1 Nk hk
λi e + [ẍi + ẋi (γ˙0 − γ̇)]e−2γ = −κ2 A(µ) hi −
µ=1
D−2
" n #
X (µ)
× exp Ni hi xi − 2γ0 . (7.1234)
i=1
Here
Gij = Ni δij − Ni Nj (7.1235)
are the components of the minisuperspace metric,
à n m̄
" n #!
1 X i
X X (µ)
V = e2γ − λi Ni e−2x + κ2 A(µ) exp Ni hi xi − 2γ0 . (7.1236)
2 i=1 µ=1 i=1
The dependence on the densities ρ(µ) in (7.358),(7.359) has been canceled according to the relations
(7.357).
It is not difficult to verify that after the gauge fixing γ = F (x1 , . . . , xn ) the equations of motion
(7.359) may be considered as the Lagrange-Euler equations obtained from the Lagrangian
à n !
1 X
L = eγ0 −γ Gij ẋi ẋj − V (7.1237)
2 i,j=1
The form is nongenerated and the inverse matrix to (Gij ) has the components
δ ij 1
Gij = + . (7.1242)
Ni 2 − D
The form < ., . > endows the space Rn with the metric, which signature is (−, +, ..., +) [54], [55].
By the usual way we may introduce the covariant components of vectors. For the vectors vk and uµ
we have
X n
δi (k) j
= −2 k , vi =
i
v(k) Gij v(k) = 2(Ni − δik ), (7.1243)
Nk i=1
Pn (µ) Xn
(µ) k=1 Nk hk (µ) (µ)
ui(µ) = hi − , ui = Gij uj(µ) = Ni hi . (7.1244)
D−2 i=1
The values of < vk , vi >, < vk , uµ > and < uµ , uν > are presented in Table 7.4.1.
< ., . > vj uν
δ (ν)
vi 4( Niji − 1) −2hi
TABLE 7.4.I. Values of the bilinear form < ., . > for the vectors vi and uµ , induced by curvature
of the space Mi and µ-th component of the perfect fluid correspondingly.
A vector y ∈ Rn is called time-like, space-like or isotropic, if < y, y > has negative, positive or
null values correspondingly. Vectors y and z are called orthogonal if < y, z >= 0. It should be
noted that the curvature induced vector vi is always time-like, while the perfect fluid induced vector
uµ admits any value of < uµ , uµ > (see Table 7.4.1).
Using the notation < ., . > and the vectors (7.363)-(7.365), we may write the zero-energy
constraint (2.12) and the Lagrangian (2.16) in the form
1
E= < ẋ, ẋ > +V = 0, (7.1245)
2µ ¶
γ0 −γ 1
L=e < ẋ, ẋ > −V , (7.1246)
2
where " #
n m̄
1X i X
V = e2(γ−γ0 ) − λ Ni e<vi ,x> + κ2 A(µ) e<uµ ,x> . (7.1247)
2 i=1 µ=1
CBPF-MO-003/02 178
It is obviously from (7.372) that the term induced in the potential by the non-Ricci flat space Mi
is similar to the term induced by µ-component of the perfect fluid. Due to this fact the non-zero
curvature of the manifold Mi may be also called a component and now we use the notion of the
component in such new sense. Further we employ the so called harmonic time gauge, which implies
n
X
γ(t) = γ0 = Ni x i . (7.1248)
i=1
From the mathematical viewpoint the problem consist in integrability of the system with n ≥ 2
degrees of freedom, described by the Lagrangian of the form
Xm
1
L = < ẋ, ẋ > − a(µ) e<bµ ,x> , (7.1249)
2 µ=1
where x, bµ ∈ Rn . In (7.374) m denotes the total number of the components including the curvatures
and the perfect fluid components. It should be noted that the kinetic term < ẋ, ẋ > is not positively
definite bilinear form as it usually takes place in classical mechanics. Due to the pseudo-Euclidean
signature (−, +, ..., +) of the form < ., . > such systems may be called pseudo-Euclidean Toda-like
systems as the potential like that given in (7.374) defines well known in classical mechanics Toda
lattices [226]. In the papers [73], [61], [72] the following classes of the integrable pseudo-Euclidean
Toda-like systems have been found
1. m = 0. This case corresponds to the vacuum multidimensional cosmological model on the
manifold M = R × M1 × . . . × Mn with all Ricci-flat spaces Mi . The corresponding metric is
a multidimensional generalization of the well-known Kasner solution [61].
2. m=1, the vector b1 is arbitrary. The metrics for this 1-component case were obtained in [72].
This integrable class may be enlarged by the addition of the new components inducing the
vectors collinear to the vector b1 .
3. m ≥ 2, n = 2, bµ = b + Cµ b0 , where b is an arbitrary vector and b0 is an arbitrary isotropic
vector, Cµ =const. This class was integrated in [72] only under the zero energy constraint.
4. m ≥ 2, the vectors b1 , . . . , bm are linear independent and satisfy the conditions < bµ , bν >= 0
for µ 6= ν. This integrable class may be enlarged by the addition of the new components
inducing the vectors collinear to one from the orthogonal set b1 , . . . , bm . The corresponding
cosmological models are studied in [73].
5. m ≥ 2, the vectors b1 , . . . , bm are space-like and may be interpreted as a set of admissible
roots [189] of a simple complex Lie’s algebra G. In this case the pseudo-Euclidean Toda-like
system is trivially reducible to the Toda lattice associated with the Lie algebra G [226]. For
G = A2 ≡ sl(3, C) the metric of the corresponding cosmological model was explicitly written
in [?].
We consider only 2-component (m = 2) pseudo-Euclidean Toda-like systems with 2 degrees of
freedom (n = 2) under the zero energy constraint. The corresponding multidimensional cosmological
models are 2-factor spaces, i.e.
M = R × M1 × M2 (7.1250)
and admit the following combinations of the components: curvature of M1 and curvature of M2
(vacuum models); curvature of M1 or M2 and 1-component perfect fluid; 2-component perfect fluid
CBPF-MO-003/02 179
in Ricci-flat spaces M1 and M2 . In our recent paper [114] we have integrated the vacuum model of
the type (7.375) with 2 curvatures for the dimensions (dimM1 ,dimM2 )=(6,3),(8,2),(5,5). Now we
develop more general procedure useful for any combination of the 2 components.
the Lagrangian (7.374) and the corresponding zero-energy constraint look as follows
¡ ¢
L = 2β ż 2 − ẏ 2 − V (z, y), (7.1259)
¡ ¢
E = 2β ż 2 − ẏ 2 + V (z, y) = 0, (7.1260)
CBPF-MO-003/02 180
It should be mentioned that using of a basis in the form (7.376) provides the factorization of the
potential (7.386) with respect to the coordinates of the vector x(t) (the additional linear trans-
formation (7.382),(7.383) does not matter in this situation). Such factorization of the potential is
essential under the developing of the following procedure proposed in [60]. Using the equation of
motion following from the Lagrangian (7.384)
1 ¡ £ ¤ £ ¤ ¢
z̈ = − V0 e2αβy β1 sgn a(1) e2β1 z + β2 sgn a(2) e2β2 z , (7.1265)
2β
α
ÿ = V (z, y), (7.1266)
4
the zero-energy condition (7.385) written in the form
1 V (z, y) 1 V (z, y)
ż 2 = 2
= (7.1267)
2β (ẏ/ż) − 1 2β (dy/dz)2 − 1
and the relation
d2 y ÿ − z̈ dy
dz
2
= 2
(7.1268)
dz ż
we obtain the following second-order ordinary differential equation
"µ ¶ #½ µ ¶ ¾
2
d2 y dy 1 e2z − ε dy
= −1 β1 + β2 + 2z + αβ , (7.1269)
dz 2 dz 2 e + ε dz
where
£ ¤
ε = sgn a(1) a(2) . (7.1270)
We notice that due to the factorization of the potential the right side of the equation (7.394) does
not contain y, so, in fact, the equation is the first-order one with respect to dy/dz.
This procedure is valid for the solutions such that ż 6≡ 0. Under the zero energy constraint the
solutions of (7.390),(7.391) with ż ≡ 0 gives the following vector x(t)
x(t) = p ln |t − t0 | + q, (7.1271)
We note that the exceptional solution (7.396) exists only if the inequalities in (7.398) are satisfied.
It should be mentioned, that the set of the equations (7.385),(7.390),(7.391) does not admit static
solutions ż = ẏ ≡ 0 due to the condition (7.378). The solutions with ż = ±ẏ are also impossible,
so using the relation (7.392) we do not lose any solutions of the set (7.385),(7.390),(7.391) except,
possibly, the solution (7.396).
Let us suppose that one is able to obtain the general solution of the equation (7.394) in the
parametrical form z = z(τ ), y = y(τ ), where τ is a parameter. Then using (7.381)-(7.383) we
obtain the vector x as the function of the parameter τ
" s¯ ¯#
2y(τ ) ¯ a(2) ¯
x(τ ) = (−β2 b1 + β1 b2 ) + 2β z(τ ) − ln ¯¯ (1) ¯¯ (b2 − b1 ). (7.1274)
α a
We recall that coordinates of the vector x(τ ) in the canonical basis are the logarithms of the scale
factors for the spaces M1 , M2 . The relation between the harmonic time t and the parameter τ may
be always derived by integration of the zero-energy constraint written in the form of the separable
equation
¡ dy ¢2 ¡ dz ¢2
− dτ
dt2 = 2β dτ dτ 2 . (7.1275)
V (z(τ ), y(τ ))
Thus the problem of the integrability by quadrature of the pseudo-Euclidean Toda-like systems with
2 degrees of freedom under the zero-energy constraint is reduced to the integrability of the equation
(7.394).
For dy/dz the equation (7.394) represents the first-order nonlinear ordinary differential equation.
Its right side is third-order polynom (with the coefficients depending on z) with respect to the
dy/dz. An equation of such type is called Abel’s equation (see, for instance [398],[397]). There are
no methods to integrate arbitrary Abel’s equation, however the equation (7.394) may be integrated
for some values of the parameters αβ and β1 + β2 . First of all let us notice that the equation
(7.394) has the partial integrals y ± z =const, which make the relation (7.392) singular and as
was already mentioned are not partial integrals of the set (7.385),(7.390),(7.391). Existence of this
partial solution of the Abel equation (7.394) allows to find the following nontrivial transformation
X dY
e2z = −ε , (7.1276)
· Y dX ¯ ¯ ¸
¯Y ¯
y = δ z + ln ¯¯ ¯¯ + ln C , δ = ±1, C > 0, (7.1277)
X
which reduces the Abel equation (7.394) to the so called generalized Emden-Fowler equation
µ ¶l
d2 Y n m dY
=X Y , (7.1278)
dX 2 dX
where the constant parameters n, m and l read
p
1 −2u11 − u22 + 3u12 − δ u212 − u11 u22
n = (β1 + β2 − 2δαβ − 3) = , (7.1279)
2 u11 + u22 − 2u12
p
1 −u11 − 2u22 + 3u12 + δ u212 − u11 u22
m =− (β1 + β2 − 2δαβ + 3) = , (7.1280)
2 u11 + u22 − 2u12
p
1 2u11 + u22 − 3u12 − δ u212 − u11 u22
l =− (β1 + β2 + 2δαβ − 3) = . (7.1281)
2 u11 + u22 − 2u12
CBPF-MO-003/02 182
For our models the parameters in the generalized Emden-Fawler equation are not independent. It
follows from (7.404),(7.405) that
n + m = −3. (7.1282)
In the special case l = 0 the equation (7.403) is known as the Emden-Fowler equation.
If the parameters l and m given by (7.406),(7.405) are such that l = 0, m 6= 1 there exists one
more transformation
2 X dY
1 + εe2z = − , (7.1283)
· m − 1 Y dX ¸
1
y=δ z− ln Y 2 + C , δ = ±1, C ∈ R, (7.1284)
m−1
which reduces the Abel equation (7.394) to the following integrable Emden-Fowler equation
d2 Y m+3
=Y m−1 . (7.1285)
dX 2
There are no methods for integrating of the generalized Emden-Fowler equation with arbitrary inde-
pendent parameters n, m and l. However, the discrete-group methods recently developed by Zaitsev
and Polyanin [397] allows to integrate by quadrature 3 two-parametrical classes, 11 one-parametrical
classes and about 90 separated points in the parametrical space (n, m, l) of the generalized Emden-
Fowler equation. For instance, the two-parametrical integrable classes arise when m and l are
arbitrary and n = 0 or when n and l are arbitrary and m = 0. The one-parametrical class with
l = 0 and n + m = −3 is also integrable by quadrature.
Let us suppose that the two components of the 2-factor spaces cosmological model under consid-
eration induce such vectors b1 and b2 that the corresponding to the model generalized Emden-Fowler
equation (3.28) with the parameters defined by (7.377),(7.404)-(7.405) is integrable in the paramet-
rical form X = X(τ ), Y = Y (τ ), where τ is a parameter. Then, using the parameter τ as the new
time coordinate we obtain by the formulas (7.401),(7.402),(7.399),(7.400) the following final result
for the metric (7.348+)
g = −f 2 (τ )[a1 (τ )]2N1 [a2 (τ )]2N2 dτ ⊗ dτ + [a1 (τ )]2 g (1) + [a2 (τ )]2 g (2) , (7.1286)
where we denoted
s · ¸l
2|β| n+l Y 0 (τ ) [X 0 (τ )]2
f (τ ) = C , (7.1287)
V0 X 0 (τ ) X(τ )Y (τ )Y 0 (τ )
(¯ ¯(2−l)bi −(1−l)bi ¯ ¯(2+n)bi −(1+n)bi ) l+n
2β
¯ 0
Y (τ ) ¯ ¯ ¯
¯ Y (τ ) ¯
(1) (2) (1) (2)
ai (τ ) ≡ ex (τ ) = eγ ¯¯ 0 ¯¯
i i
¯ X(τ ) ¯ . (7.1288)
X (τ )
We recall that bi(µ) are coordinates of the vector bµ in the canonical basis. In the special case
l = 0 one may also use by the similar manner the transformation (7.408),(7.409) and the result of
CBPF-MO-003/02 183
integration of the equation (7.410) to write the metric. This transformation was used in [114] for
integrating of the models with two curvatures.
Thus the method described allows to integrate the cosmological models if the corresponding
generalized Emden-Fowler equation is integrable. Note that if the model with some vectors b1 and b2
is integrable by such manner then any model with the vectors αb1 and αb2 (α is an arbitrary non-zero
constant) is also integrable as the parameters n, m and l do not change under such transformation
of the vectors. Taking into account the classes 1-4 (the class 5 does not arise for n = 2) and the
additional to them class, which may be integrated by the method described, we obtain the quite
large variety of the integrable 2-factor spaces cosmological models with 2 components.
1-st component of
(1) (1)
the perfect fluid h1 h2
2-nd component of
(2) (2)
the perfect fluid h1 h2
Let us consider the models of the type represented in Table 2 leading to the generalized Emden-
Fowler equation (7.403) with
l = 0. (7.1291)
Due to the relation (7.407) arising for our models the equation is integrable for arbitrary parameter
m and its exact solution has been written by Zaitsev and Polyanin [398]. It is worth to mention that
other 2 integrable classes of the generalized Emden-Fowler equation (7.403), arising when n = 0
or m = 0, describe the same cosmological models. It follows easily from (7.404),(7.406) that if the
model is such that l = 0 for δ = 1 (or δ = −1) then n = 0 for δ = −1 (correspondingly, δ = 1). It
is easy to see also from (7.405),(7.406) that the condition l = 0 transforms to the condition m = 0
under the inverse numbering of the components. Thus, from 3 integrable classes, arising for n = 0,
m = 0 and l = 0, correspondingly, of the generalized Emden-Fowler equation (7.403) for our models
it is enough to study any one from them, let it be the class with l = 0. In this case the equation
(7.403) has the form
d2 Y
2
= X −m−3 Y m , m = −β1 − β2 . (7.1292)
dX
By the following transformation
τ 1
Y = , X= (7.1293)
ξ ξ
it reduces to the equation
d2 τ
= τ m, (7.1294)
dξ 2
which is easily integrable. Then the general solution of the equation (7.417) has the form
τ 1
Y =± , X=± , (7.1295)
F (τ ) F (τ )
where
Z · ¸−1/2
2
F (τ ) = τ m+1 + C1 dτ + C2 , m 6= −1, (7.1296)
m+1
Z
= [2 ln |τ | + C1 ]−1/2 dτ + C2 , m = −1. (7.1297)
We suppose that the both components of the perfect fluid have the positive mass-energy
£ (1) (2) ¤ densities
(µ) 2 (µ)
given by (7.357). It means a = κ A > 0 for µ = 1, 2, so ε = sgn a a = 1. Then taking
into account the formula (7.401), we must consider the general solution (7.420) on such interval of
the variable τ where
X(τ )Y 0 (τ ) F (τ )
G(τ ) ≡ − 0
= − 1 > 0. (7.1298)
Y (τ )X (τ ) τ F 0 (τ )
Finally using the results of 7.4.3 we obtain the following exact solution for the cosmological
model represented by Table 7.4.2 in the special case l = 0:
the metric is given by the formula (7.411), where
· 0 ¸2
2 2|β|C −2δαβ F (τ )
f (τ ) = β −β
, (7.1299)
[A(1) ] 2 [A(2) ] 1 G(τ )τ 2
the scale factors of the spaces M1 , M2 have the form
· (1) ¸β [ui(2) −ui(1) ] n oδ/α
A ui(2) −2ui(1) £ 2 2 ¤β1 ui(2) −β2 ui(1)
ai (τ ) = [G(τ )] C τ , (7.1300)
A(2)
CBPF-MO-003/02 185
The functions F (τ ) and G(τ ) are defined in (7.421)-(7.423); components ui(µ) of the vectors in the
canonical basis are given in (7.369); the parameters α, β, β1 , β2 are defined in (7.387),(7.388); the
(µ)
values uµν =< uµ , uν > for µ, ν = 1, 2 may be calculated from the parameters Ni , hi by the formula
given in Table 7.4.1.
The following relation is valid for the densities (7.426)-(7.427)
TABLE 7.4.3. Examples of the integrable models for dimensions N1 = 3 and N2 = 6. The
corresponding exact solutions are given by the formulas (7.427)-(7.427).
upon which the Friedman-Robertson-Walker (FRW) world model is based on. It is of interest to
study processes in many dimensions when the perfect fluid is replaced by the viscous on.
Herein, we assume that the cosmic fluid (the source of the gravitational field especially at
early stages) is viscous, which might simulate high energy physics processes (such as the particles
creation). The effects related to viscosity in 4-dimensional Universe were studied through different
viewpoints (see e.g., [409-411, 412, 416-423]). Before developing the multidimensional model let
us briefly discuss (extensive review of the subject was given by Gron [415]) the main trends in
4-dimensional cosmology with viscous fluid as a source.
First, Misner [418] considered neutrino viscosity as a mechanism of reducing the anisotropy in the
Early Universe. Stewart [424] and Collins and Stewart [413] proved that it is possible only if initial
anisotropies are small enough. Another series of papers which concerns the production of entropy in
the viscous Universe was started by Weinberg [426]. Both isotropization and production of entropy
during lepton era in models of Bianchi types I,V were considered by Klimek [416]. Caderni and
Fabbri [412] calculated coefficients of shear and bulk viscosity in plasma and lepton eras within the
model of Bianchi type I. The next approach is connected with obtaining singularity free viscous
solutions. The first nonsingular solution was obtained by Murphy [420] within the flat FRW model
with fluid possessing a bulk viscosity. However, Belinsky and Khalatnikov [409, 410] showed that
this solution corresponds to the very peculiar choice of parameters and is unstable with respect
to the anisotropy perturbations. Other nonsingular solutions with bulk viscosity were obtained by
Novello and Araujo [421], Romero [423], Oliveira and Salim [422].
The crucial feature of each viscous cosmological model is assuming of the so called ”second
equation of state”, which provides us with the viscosity coefficients dependence on time. Further
we denote by ζ and η the bulk and shear viscosity coefficients, correspondingly. Murphy [420]
integrated the 4-dimensional flat FRW model with bulk viscosity by assuming ζ ∼ ρ (as second
equation of state), where ρ is the density of the viscous fluid. Belinsky and Khalatnikov [409, 410]
studied the behavior of this model as well as homogeneous anisotropic models of Bianchi types I
and IX by means of qualitative methods with more general second equations of state ζ, η ∼ ρν ,
where ν is constant. Lukacs [417] integrated the homogeneous and isotropic 4-dimensional model
with a viscous pressureless fluid and a second equation of state given by ζ ∼ [scale factor]−1 .
A curvature-dependent bulk viscosity was studied in multidimensional cosmology by Wolf [427].
Recently, Motta and Tomimura [419] studied a 4-dimensional inhomogeneous cosmology with a
bulk viscosity coefficient which depends on the metric. In our papers [116] exact solutions in
multidimensional models with bulk viscosity were obtained and their properties were studied for
the following type of the second equation of state: ζ ∼ [ volume of the Universe]−1 .
The aim of the present investigation is to integrate the Einstein equations for a
multidimensional cosmological model formed by a chain of Ricci-flat spaces and a cosmic fluid pos-
sessing both shear and bulk viscosity. The second equations of state are chosen in the following form
of metrical dependence of the bulk and shear viscosity coefficients:
−1
ζ, η ∼ [volume of the Universe] .
In 7.5.2 we describe the model and get basic equations. To integrate them, we develop some
vector formalism proposed in [14, 114]. Thermodynamical concepts in multidimensional cosmologies
are defined in 7.5.3, where a formula which provides us with the variation rate of the entropy is
derived. The equations of motion for the special set of parameters in the first and the second
equations of state are integrated in 7.5.4. Exact solutions are given in a Kasner-like form, their
physical properties are investigated in 7.5.5, where the process of dynamical compactification and
the entropy production are also defined.
CBPF-MO-003/02 188
on the manifold defined in Eq. (7.1304), where ds2i is the metric of the Einstein space Mi , γ(t)
and xi (t) arePn scalar functions of the cosmic time t. The dimension of this manifold is given by
D = 1 + i=1 Ni , where Ni = dim Mi . Herein, for reason of simplicity, only Ricci-flat spaces
M1 , . . . , Mn are assumed (i.e., the components of the Ricci tensor for the metrics ds2i are zero). One
easily obtains the non-zero components of the Ricci-tensor for the metric defined in Eq. (7.1305)
(see [61]):
à n !
X
R00 = e−2γ(t) Ni (ẋi )2 + γ̈0 − γ̇ γ̇0 , (7.1306)
i=1
¡ ¢
Rnmii = e −2γ(t)
ẍ + (γ̇0 − γ̇)ẋi δnmii ,
i
(7.1307)
Pn Pn
for i = 1, . . . , n, where the indices mi and ni run from D − j=i Nj to D − j=i Nj + Ni and
P
γ0 = ni=1 Ni xi .
A viscous fluid is characterized by a density ρ, a pressure p, a bulk viscosity coefficient ζ, a shear
viscosity coefficient η, so that the (standard form of the) energy-momentum tensor reads
Tνµ = ρuµ uν + (p − ζθ)Pνµ − 2ησνµ , (7.1308)
where uµ is the D-dimensional velocity of the fluid, θ = uµ;µ denotes the scalar expansion, Pνµ = δνµ +
uµ uν is the projector on the (D−1)-dimensional space orthogonal to uµ , and σνµ = 12 (uα;β + uβ;α ) P αµ Pνβ −
(D − 1)−1 θPνµ is the traceless shear tensor (defined as usual).
By choosing the D-dimensional velocity so that uµ = δ0µ e−γ(t) (the comoving observer condition),
we obtain
θ = γ̇0 e−γ(t) , (7.1309)
(uµ uν ) = diag(−1, 0, . . . , 0), (7.1310)
(Pνµ ) = diag(0, 1, . . . , 1), (7.1311)
µ µ ¶ µ ¶ ¶
γ̇0 γ̇0
(σν ) = e diag 0, ẋ1 −
µ −γ
δlk11 , . . . , n
ẋ − δlknn , (7.1312)
D−1 D−1
where ki , li = 1, . . . , Ni for i = 1, . . . , n. The function γ(t) determines a time gauge (a harmonic
time gauge for γ(t) = γ0 and a synchronous time gauge for γ(t) = 0), see Eq. (7.1305); note that the
harmonic time t and the synchronous time ts are related by dts = exp[γ0 ]dt. By assuming anisotropy
properties for the pressure and the bulk viscosity, with respect to the whole space M1 × . . . × Mn ,
one has
(Tνµ ) = diag(−ρ, p∗1 δlk11 , . . . , p∗n δlknn ), (7.1313)
where · ¶¸
µ
γ̇0
p∗i = pi − e −γ
ζi γ̇0 + 2η ẋ − i
, (7.1314)
D−1
and pi , resp. ζi , is the pressure, resp. the bulk viscosity coefficient, in the space described by the
manifold Mi . Furthermore, we assume that the barotropic equations of state hold
pi = (1 − hi )ρ(t), (7.1315)
CBPF-MO-003/02 189
where the hi are constants. One easily shows that the form of the equation of motion (5M T0M = 0)
for a viscous fluid described by a tensor given by Eq. (7.1313), is given by
n
X
ρ̇ + Ni ẋi (ρ + p∗i ) = 0. (7.1316)
i=1
The Einstein equations Rνµ − 21 δνµ R = κ2 Tνµ , where κ2 is the gravitational constant, can be
T
written as Rνµ = κ2 (Tνµ − D−2 δνµ ). Further, by using the equations R00 − 12 δ00 R = κ2 T00 and Rnmii =
T
κ2 (Tnmi i − D−2 δkmi i ), Eqs.(7.1306,7.1307,7.1313) give the following equations of motion
n
X
Ni (ẋi )2 − γ̇02 = −2κ2 e2γ ρ, (7.1317)
i=1
· µ Pn
¶
i i 2 γ γ Nk h k k=1
ẍ + (γ̇0 − γ̇)ẋ = κ e e ρ −hi +
D−2
µ Pn ¶¸ µ ¶
k=1 Nk ζk 2 γ i γ̇0
+ γ̇0 −ζi + − 2κ e η ẋ − . (7.1318)
D−2 D−1
We use an integration procedure which is based on the n-dimensional real vector space Rn . Let
e1 , . . . , en be the canonical basis in Rn (i.e. e1 = (1, 0, . . . , 0) etc. . . ), and h, i denote a symmetrical
bilinear form defined on Rn by
It has been used as a mini-super-space metric for cosmological models (see [52, 54, 72, 87). Such a
form is non-generate and has the pseudo-Euclidean signature (−, +, . . . , +). With this in mind, a
vector y ∈ Rn is time-like, resp. space-like or isotropic, if hy, yi takes negative, resp. positive or null
values; and two vectors y and z are orthogonal if hy, zi = 0. Hereafter, we use the following vectors
x = x1 e1 + . . . + xn en , (7.1320)
Pn
Nk h k
u = u1 e1 + . . . + un en , ui = hi − k=1 , u i = Ni h i , (7.1321)
PnD − 2
Nk ζk
ξ = ξ 1 e1 + . . . + ξ n en , ξ i = ζi − k=1 , ξi = Ni ζi , (7.1322)
D−2
where covariant coordinates of the vectors are introduced by the usual way. Moreover, let us denote
ud the particular vector given by Eq. (7.446) with hi=1,...,n = 1 (it is related to dust in the whole
space, see Eq. (7.1315)). One has
−1 D−1
(ud )i = Ni , uid = , hud , ud i = − , hud , xi = γ0 . (7.1323)
D−2 D−2
Thus, using Eqs. (7.445-7.448) we rewrite the Einstein equations (7.442),(7.443) in the form
where the formal dependence on ρ in Eq. (7.450) has been canceled, according to Eq. (7.449). More-
over, Eq. (7.1316) can be written as
µ ¶
−γ 2ηud
ρ̇ + h2ud − u, ẋiρ − e 2ηhẋ, ẋi + hud , ẋihξ − , ẋi = 0. (7.1326)
hud , ud i
To integrate Eq.(7.450) one needs second equations of state, involving the bulk viscosity coefficients
ζ1 , . . . , ζn and the shear viscosity coefficient η. Let us assume that these coefficients are proportional
to exp[−γ0 ] (or inversely proportional to the volume of the Universe), i.e.
which means (from a physical viewpoint) that the viscosity decreases when the space M1 × . . . × Mn
expands. The integrability of the basic equation (provided the second equations of state) is ensured
when the vectors u, ξ, ud are either colinear or orthogonal (with respect to the mini-super-space
metric) in some combination [116-414]. Herein, we suppose that these vectors are colinear, which
means that the viscous fluid has identical properties in the internal space(s) and the external space.
Hence, all these assumptions, for the pressures and the viscosity coefficients, allow us to write
where Vi stands for a fluid volume in the space Mi , when V = V1 · . . . · Vn is a fluid volume in the
whole space, and S is an entropy in the volume V . By assuming that the baryon particle number
NB in the volume V is conserved, Eq. (7.1331) transforms to
n
X n
X
nT ṡ = ρ̇ + ρ Ni ẋi + pi Ni ẋi , (7.1332)
i=1 i=1
where s = S/NB , resp. n = NB /V , stands for the entropy per baryon, resp. the baryon number
density. Let us remind that exp[xi ] is the scale factor of the space Mi (of dimension Ni ). For a perfect
fluid (ζi = 0, η = 0), the comparison between Eqs. (7.1316,7.1332) gives the entropy conservation
( i.e., s is constant). Similarly, we obtain also the temperature (see [428]). From Eq. (7.1332) we
have µ ¶
∂ρ
= −ρNi − pi Ni = (hi − 2)Ni ρ, (j 6= i), (7.1333)
∂xi s,xj
CBPF-MO-003/02 191
where K(s) is an unknown function (which reads in term of the entropy per baryon s). Using
Eqs. (7.457),(7.459) we get
µ ¶ " n #
∂ρ X
0 i
= nT = K (s) exp (2 − hi )Ni x . (7.1335)
∂s xi i=1
For a fluid with a bulk and shear viscosity, the comparison between Eqs. (7.1316,7.1332) provides
us with µ ¶
−γ 2η
nT ṡ = e 2ηhẋ, ẋi + hud , ẋihξ − 2 ud , ẋi . (7.1337)
hud , ud i
Such a formula gives the variation rate of entropy per baryon in multidimensional cosmology on the
manifold M = R × M1 × . . . × Mn with viscosity. The entropy production can be calculated if the
temperature of the fluid is known. Herein, we suppose that the temperature is given by the perfect
fluid formula Eq. (7.461), which is valid with sufficient accuracy when effects of viscosity are small.
Hence, Eqs. (7.461),(7.462) give
µ ¶
h2ud −u,xi−γ 2ηud
ṡ = Be 2ηhẋ, ẋi + hud , ẋihξ − 2 , ẋi . (7.1338)
hud , ud i
Let us note that the basis vectors f2 , . . . , fn are space-like, since they are orthogonal to the time-like
vector ud . The vector x ∈ Rn decomposes as follows
X n
ud
x = hud , xi + hfj , xifj . (7.1342)
hud , ud i j=2
CBPF-MO-003/02 192
Hence, the basic vector equation given in Eq. (7.1339) reads in term of coordinates in such a basis
as follows ¡ ¢
hfj , ẍi + hud , ẋi − γ̇ + 2ηκ2 eγ hfj , ẋi = 0 (7.1343)
µ ¶ " n
#
hud , ẍi hud , ẋi − γ̇ 2 γ h hud , ẋi2 X
+ + ζκ e hud , ẋi = + hfj , ẋi2 , (7.1344)
hud , ud i hud , ud i 2 hud , ud i j=2
which provides us with a set of equations (for j = 2, . . . , n). For the metric dependence of the
viscosity coefficients, see Eqs. (7.454),(7.455) Eqs. (7.1343,7.1344) read
Such a set of equations is integrable for any values of the constant parameters h, η0 and ζ0 .
Let us first assume models with
h 6= 0. (7.1348)
The integration of Eq. (7.1345) gives
½
tpj + q j if η0 = 0,
hfj , xi = (7.1349)
e−η0 t pj + q j if η0 =
6 0,
where pj and q j are arbitrary constants. The Kasner-like form solution can be written in term of
vectors α, β ∈ Rn , defined as follows
n
X n
X n
X n
X
j i j
α= p fj ≡ α ei , β= q fj ≡ β i ei , (7.1350)
j=2 i=1 j=2 i=1
where αi and β i are their coordinates in the canonical basis e1 , . . . , en . By using the orthogonality
conditions, we obtain
n
X n
X
i
hα, ud i = α Ni = 0, hβ, ud i = β i Ni = 0, (7.1351)
i=1 i=1
n
X n
X
¡ i ¢2 ¡ j ¢2
hα, αi = α Ni = p ≥0 (7.1352)
i=1 j=2
n
X n
X
¡ i ¢2 ¡ j ¢2
hβ, βi = β Ni = q ≥ 0, (7.1353)
i=1 j=2
where the constants αi and β i may be called Kasner-like parameters, because of the existence of
these constraints. By using Eqs. (7.1342,7.1349,7.1350), we obtain
ud
x = hud , xi + a(t)α + β, (7.1354)
hud , ud i
CBPF-MO-003/02 193
see [408], where I|ν| (τ ), resp. K|ν| (τ ), is the related modified Bessel function, resp. Mac-Donald
function.
Finally, the results of Eqs. (7.1345,7.1346) integration for various constants ζ0 and η0 can be
presented in Tab. 7.5.1
Table 7.5.1
ζ0 = 0 ζ0 6= 0
η0 6= 0 Solution III with hα, αi 6= 0 and Solution III with hα, αi 6= 0 and
Solution I Solution II with hα, αi = 0
CBPF-MO-003/02 194
• Solution I :
i i
ex = eβ |C1 + C2 t|−2/[h(D−1)] . (7.1363)
• Solution II :
¯ ¯−2/[h(D−1)]
i i i ¯ ¯
ex = eα t+β ¯C1 e−(Ã−ζ̃0 )ht/2 + C2 e(Ã+ζ̃0 )ht/2 ¯ , (7.1364)
where
s µ ¶2 q
D − 1 ζ0 D−1 D − 1 ζ0
ζ̃0 = , Ã = hα, αi + = A2 + ζ̃02 . (7.1365)
D−2 h D−2 D−2 h
• Solution III :
¡ ¢¡ ¯ ¯¢−2/[h(D−1)]
= exp αi e−η0 t + β i τ −ν ¯C1 I|ν| (τ ) + C2 K|ν| (τ )¯
i
ex , (7.1366)
where the variable τ > 0 is given in Eq. (7.1359), the constants A in Eq. (7.1357) and ν in
Eq. (7.1361).
In formulas (7.488),(7.490),(7.491) Ci=1,2 are integration constants such that C12 + C22 > 0 and
i = 1, . . . , n. The Kasner-like parameters αi and β i obey the relations given in Eq. (7.476).
The set of equations given in Eqs. (7.1345,7.1346) is easily integrable in the case
h = 0, (7.1367)
which, with the barotropic equation of state in mind, relates to Zeldovich or stiff matter. The
results are given as follows :
• Solution IV : for i = 1, . . . , n
½
£ ¤ exp (C
e xi i
= exp α a(t) + β × i ¡ 1 + C2 t) ¡ D−1 ¢¢ if ζ0 = 0 (7.1368)
exp C1 + C2 exp ζ0 D−2 t if ζ0 6= 0,
where Ci=1,2 are arbitrary constants, and the function a(t) is given in Eq. (7.1355).
7.5.5 Discussion
Let us remind the multidimensional generalization of the well-known Kasner solution [72], it reads
(for the synchronous time ts ) as follows
n
X
2 i
ds = −dt2s + Ai t2ε 2
s dsi . (7.1369)
i=1
Such a metric describes the evolution of a vacuum model defined on the manifold R ×M1 ×. . .×Mn ,
where the Mi are Ricci-flat spaces of dimension Ni with the metric ds2i , Ai are arbitrary constants
and εi are the Kasner parameters, which satisfy the relations
n
X n
X
Ni εi = 1, Ni (εi )2 = 1. (7.1370)
i=1 i=1
CBPF-MO-003/02 195
The εi and the Kasner-like parameters αi (used in the above formulas for the exact solutions) are
related as
αi 1
εi = ± + , hα, αi 6= 0. (7.1371)
A D−1
By using Eq. (7.1354) (i.e., a general decomposition of the vector x ∈ Rn ) and the result of Sec.4,
see Eq. (7.463), we obtain the variation rate of entropy
B −hγ0 ¡ 2 2
¢
ṡ = e ζ0 γ̇ 0 + η 0 hα, αiȧ (t) , (7.1372)
κ2
which shows that the entropy increases when ζ0 dt > 0 and η0 dt > 0. Further we assume
ζ0 ≥ 0, η0 ≥ 0, (7.1373)
so harmonic time t (as well as synchronous time ts ) increases during the evolution.
One easily shows that the weak energy condition Tνµ v ν vµ ≥ 0, for any D-dimensional non space-
like vector v ν (and thus as well as for the 4-dimensional case), applied to the stress-energy tensor
given in Eq. (7.1313), can be written as inequalities
where ρ, resp. p∗i , is the density, resp. the effective pressure, of the fluid.
The dominant energy condition Tνµ v ν vµ ≥ 0 and Tµν Tλµ vν v λ ≤ 0, for any non space-like vector
v µ , applied to the stress-energy tensor given in Eq. (7.1313), reduces to inequalities defined in
Eq. (7.1374), with the following additional condition
ρ − p∗i ≥ 0, (i = 1, . . . , n) (7.1375)
Notice that due to the weak energy condition the following restriction on the constant h (taken
from the barotropic equation of state (7.440)) arises in the nonviscous case: h ≤ 2. The dominant
energy condition for the nonviscous stress-energy tensor implies: 0 ≤ h ≤ 2.
It is important to note that Solution I and Solutions II,IV for zero Kasner-like parameters
i
(α = 0, i = 1, . . . , n) are isotropic, since the spaces M1 , . . . , Mn have identical scale factors.
Further we discuss the solutions obtained, which are of interest within multidimensional or
4-dimensional cosmology.
Non Viscous Models
For a better understanding of the viscosity effect on the dynamics, we first outline the properties
of non viscous models.
a) The isotropic non viscous model is described by Solution I (h 6= 0) and Solution IV (h = 0)
with αi = 0 (for i = 1, . . . , n), which represents the multidimensional generalization of the flat FRW
model. It is the steady-state model
· ¸
xi C2
e ∼ exp − ts , ρ = const (7.1376)
D−1
for h = 2 and shows a power-law behavior
i
ex ∼ t2/[(2−h)(D−1)]
s , ρ ∼ t−2
s (7.1377)
for h 6= 2, where ts is the synchronous time (stationary solution with zero density is also possible);
let us call it as Friedman-like behavior.
CBPF-MO-003/02 196
b) The anisotropic non viscous model for h ∈ (0, 2] is described by Solution II with ζ0 = 0 and
hα, αi 6= 0. One easily shows that the integration constants Ci=1,2 have to satisfy the condition
C1 C2 < 0, otherwise ρ ≤ 0, ∀t. By using this condition, we obtain from Eq. (4.26)
Models with bulk viscosity Let us now investigate the models with bulk viscosity.
c) The viscous isotropic model shows interesting features. The shear viscosity is not significant
in this case and the model is described for h 6= 0 by Solution II with αi = 0 for i = 1, . . . , n. If
C1 C2 < 0 and h > 0 then the solution can be written as follows : for i = 1, . . . , n
µ · ¸¶−2/[h(D−1)]
xi D−1
e = Ri 1 − exp ζ0 (t − t0 ) , (7.1380)
D−2
µ · ¸¶
∗ D−1
pi = 1 − h exp − ζ0 (t − t0 ) ρ, (7.1381)
D−2
n · ¸
D − 1 2ζ02 Y −2Ni D−1
ρ = R exp 2 ζ0 (t − t0 )
D − 2 κ2 h2 i=1 i D−2
µ · ¸¶2(2−h)/h]
D−1
× 1 − exp ζ0 (t − t0 ) , (7.1382)
D−2
CBPF-MO-003/02 197
n · ¸
D − 1 2Bζ02 Y −hNi D−1
s = R exp 2 ζ0 (t − t0 ) + s(−∞), (7.1383)
D − 2 κ2 h2 i=1 i D−2
¯ ¯−2/[h(D−1)] ·µ ¶ ¸
xi ¯ ¯ i ζ0
e = Ri ¯sinh[Ãh(t − t0 )/2]¯ exp α − t , (7.1385)
h(D − 2)
µ ¶
h cosh2 ã
p∗i = 1− ρ, (7.1386)
1 + sinh ã sinh[Ãh(t − t0 ) + ã]
hα, αi Y −2Ni ¯¯ ¯2(2−h)/h
n
¯
ρ = R ¯sinh[ Ãh(t − t0 )/2]¯ exp[2ζ̃0 t]
2κ2 i=1 i
³ ´
× 1 + sinh ã sinh[Ãh(t − t0 ) + ã] , (7.1387)
n · µ ¶ ¸
Bζ0 A2 Y −hNi D−1 2
ṡ = R exp ζ0 1 − t sinh2 [Ãh(t − t0 )]
κ2 i=1 i D−2 h
× sinh2 [Ãh(t − t0 ) + ã], (7.1388)
where we use independent constants t0 , R1 , . . . , Rn defined by
C C
C1 = − exp[Ãht0 /2], C2 = exp[−Ãht0 /2], (7.1389)
2 2
CBPF-MO-003/02 198
i i
ex = Ri tαs /(Ã+|ζ̃0 |)+1/(D−1) , ts > 0, (7.1393)
D−2 |ζ̃0 |
p∗i = ρ = t−2 , (7.1394)
D − 1 κ2 (Ã + |ζ̃0 |) s
(D − 2)B|ζ0 |(Ã + |ζ̃0 |) −h
s = ts + s(0), (7.1395)
(D − 1)κ2
CBPF-MO-003/02 199
It is evident from Eq. (7.518) that αi /(Ã + |ζ̃0 |) + 1/(D − 1) → εi (Kasner parameter) as |ζ̃0 |/A → 0.
Therefore, if the parameter |ζ̃0 |/A is small enough, then the model describes the contraction of a
part of the spaces M1 , . . . , Mn and the expansion of another part. According to Eq. (7.520), the
solution describes the unbounded entropy production as h < 0.
We note, that the anisotropic viscous model described by Solution IV (h = 0) does not satisfy
the weak energy condition as the density has negative values at some stage of the evolution.
it can be proved that the density given has negative values at some interval of time if C2 = 0. If
C2 6= 0 then the Solution III can be written as follows: for i = 1, . . . , n
i ¡ ¢−2/[h(D−1)] £ ¤
ex = Ri τ −ν |CIν (τ ) + Kν (τ )| exp αi e−η0 t , (7.1398)
·µ ¶ ¸
2η0 τ CIν+1 (τ ) − Kν+1 (τ ) 1
ẋi = +1 − εi , (7.1399)
h CIν (τ ) + Kν (τ ) D−1
i CIν+1 (τ )−Kν+1 (τ )
h (D − 1)ε − 1 − ν CIν (τ )+Kν (τ )
p∗i = 1 − h + ³ ´2 ρ, (7.1400)
τ CIν+1 (τ )−Kν+1 (τ )
CIν (τ )+Kν (τ )
−1
n
2η 2 D − 2 Y −2Ni 2(1−2 ν ) 4
ρ = 2 02 Ri τ h |CI (τ ) + K (τ )| h
ν ν
h κ D − 1 i=1
"µ ¶2 #
CIν+1 (τ ) − Kν+1 (τ )
× −1 , (7.1401)
CIν (τ ) + Kν (τ )
n
4η03 B D − 1 Y −hNi 2(1−ν)
ṡ = R τ
h2 κ2 D − 2 i=1 i
¡ ¢
× ν[CIν+1 (τ ) − Kν+1 (τ )]2 + [CIν (τ ) + Kν (τ )]2 , (7.1402)
where the independent constants R1 , . . . , Rn , C = C1 /C2 are used. The variable τ was determined
in Eq. (7.1359). Also we introduced Kasner parameters εi by
αi 1
εi = sgn[h] + . (7.1403)
A D−1
By using the properties given in Eq. (7.492) and the asymptotical behavior of the modified Bessel
functions [408], one may prove that the density given in Eq. (7.526) has no negative values during
the evolution only for :
CBPF-MO-003/02 200
1. partial solution with C = 0 on the interval (−∞, +∞) of the harmonic time;
2. solution with C < 0 on the interval (t0 (τ0 ), +∞), where τ0 is the root of the equation
We note that the asymptotical behavior of the solution with C < 0 as t → +∞ is given by the
partial solution with C = 0.
Let us now investigate the models having a singularity at the beginning of the evolution. Such
solutions arise when h < 0. In this case for the solutions with C < 0 one obtain the following
correspondences (−τ ) ∈ (−τ0 , 0) ⇔ t ∈ (t0 (τo ), +∞) ⇔ ts ∈ (0, +∞). The singularity at ts = 0
is of Friedman type. The shear viscosity leads to the isotropization at the final stage of evolution.
1/(D−1)
We obtain from Eq. (7.523-7.526) exp[xi ] ∼ ts and p∗i = ρ ∼ t−2
s as ts → +∞, i.e. the model
describes the isotropic Friedman-like expansion corresponding to Zeldovich matter (h = 0). The
anisotropic behavior is possible on some interval of time after the Friedman-like behavior. One
can prove that if the constant |C| 6= 0 is small enough then on some interval the sign of the
Hubble parameter ẋi coincides with the sign of the Kasner parameter εi (see Eq. (7.524) for h < 0).
However, such regime is possible only on limited interval of time, because the final stage of the
evolution exhibits the isotropic expansion due to the shear viscosity.
shear and bulk viscosity coefficients on time, in some special form of their metric dependence. The
exact solutions for both isotropic and anisotropic case are obtained in subsection 7.6.3, where their
dynamical properties are discussed.
M = R × M1 × . . . × Mn , (7.1406)
where ds2i is the metric of the Ricci-flat factor space Mi of dimension di , γ(t) and xi (t) are scalar
functions of the cosmic time t. ai ≡ exp[xi ] is the scale factor of the space Mi and the function γ(t)
determines a time gauge. The synchronous time ts is defined by the equation dts = exp[γ(t)]dt.
We consider a source of gravitational field in the form of 2-component cosmic fluid. The first
one is a perfect fluid with a density ρ(1) (t) and a pressure p(1) (t). The second component supposed
to be a viscous fluid. It is characterized by a density ρ(2) (t), a pressure p(2) (t), a bulk viscosity
coefficient ζ(t) and a shear viscosity coefficient η(t). The overall energy-momentum tensor of the
cosmic fluid reads
¡ ¢ ¡ ¢
Tνµ = ρ(1) + ρ(2) uµ uν + p(1) + p(2) − ζθ Pνµ − 2ησνµ , (7.1407)
where uµ is the D-dimensional velocity of the fluid, θ = uµ;µ denotes the scalar expansion, Pνµ = δνµ +
uµ uν is the projector on the (D−1)-dimensional space orthogonal to uµ , σνµ = 21 (uα;β + uβ;α ) P αµ Pνβ −
(D − 1)−1 θPνµ is the traceless shear tensor and µ, ν = 0, 1, . . . , D − 1.
By assuming the comoving observer condition uµ = δ0µ e−γ(t) , the overall energy-momentum
tensor may be written as
Tνµ = Tνµ(1) + Tνµ(2) , (7.1408)
where
¡ ¢
Tνµ(1) = diag(−ρ(1) , p(1) , . . . , p(1) ), (7.1409)
¡ µ(2) ¢ (2)
Tν = diag(−ρ(2) , p̃1 δlk11 , . . . , p̃(2) kn
n δln ), (7.1410)
(2)
ki , li = 1, . . . , di for i = 1, . . . , n. Here p̃i denotes the effective pressure including the dissipative
contribution of the viscous fluid in the factor space described by the manifold Mi . It reads
· µ ¶¸
(2) (2) −γ i γ̇0
p̃i = p − e ζ γ̇0 + 2η ẋ − , (7.1411)
D−1
where n
X
γ0 = di xi . (7.1412)
i=1
i = 1, . . . , n.
µ(2)
The energy conservation law 5µ T0 = 0 for a viscous fluid described by a tensor given by
equation (7.1410) reads
Xn ³ ´
(2) i (2) (2)
ρ̇ + di ẋ ρ + p̃i = 0. (7.1416)
i=1
Taking into account equation (7.1413), ones easily integrates equation (7.542). The result is
(1) −2)γ
ρ(1) = Ae(h 0
, (7.1418)
where A is a positive constant. We note that the contribution of the perfect fluid component
with h(2) = 2 to the overall energy momentum tensor is equivalent to the presence of Λ-term with
Λ = κ2 A.
Furthermore, by using equations (7.539) and (7.543) the presence of the densities ρ(1) and ρ(2)
in equations (7.1415) can be cancelled. Thus, we obtain the main governing set of equations
à n
!
h (1)
− h (2)
(1) h (2) X
ẍi + (γ̇0 − γ̇)ẋi = κ2 A e[h −2]γ0 +2γ + γ̇02 − di (ẋi )2
D−2 2(D − 2) i=1
(7.1419)
· µ ¶¸
ζ γ̇0
+ κ2 eγ γ̇0 − 2η ẋi − . (7.1420)
D−2 D−1
Such a form is non-degenerate and has the pseudo-Euclidean signature (−, +, . . . , +) [55] . With
this in mind, a vector y ∈ Rn is timelike, spacelike or isotropic respectively, if hy, yi takes negative,
CBPF-MO-003/02 203
positive or null values respectively and two vectors y and z are orthogonal if hy, zi = 0. Hereafter,
we use the following vectors
where s = S/NB and n = NB /V stands for the entropy per baryon and the baryon number density.
The comparison between equations (7.542) and (7.554) gives the variation rate of entropy per baryon
n
X ³ ´ ·µ ¶ ¸
i (2) (2) −γ D−2 2
nT ṡ = di ẋ p − p̃i = e ζ+ 2η γ̇0 + 2ηhẋ, ẋi . (7.1430)
i=1
D−1
where ζ0 ≥ 0 and η0 are constants. When the cosmological model is isotropic, i.e. ẋi = γ̇0 /(D −
1), i = 1, . . . , n, and the shear viscosity is not significant, the expression H ≡ γ̇0 e−γ /(D − 1) is
the Hubble parameter.Then we get from equation (7.556): ζ = D−2 ζ H, i.e. the bulk viscosity
κ2 0
coefficient is linear proportional to the Hubble parameter. Such kind of the second equation of state
describes the so called ”linear dissipative regime” in the FRW world model (see, for instance, [435]).
As we study an anisotropic cosmological model, we must involve a shear viscosity as well as a bulk
one. So, we propose equations (7.556) for the anisotropic model as a simplest generalization to the
of the second equation of state describing the linear dissipative regime.
In order to integrate equation (7.551), we use the orthogonal basis
u
, f2 , . . . , fn ∈ Rn , (7.1432)
hu, ui
where the vector u was introduced by equation (7.1423) The orthogonality property reads
Let us note that the basis vectors f2 , . . . , fn are space-like, since they are orthogonal to the time-like
vector u. The vector x ∈ Rn decomposes as follows
X n
u
x = γ0 + hx, fj ifj . (7.1434)
hu, ui j=2
Hence, under the above assumptions equation (7.551) reads in the terms of coordinates in such basis
as follows
h³ (2)
´ i
γ̈0 + 1 − h 2 − ζ0 γ̇0 − γ̇ γ̇0 =
(7.1435)
D−1 h ¡ ¢ (1) (2) Pn
i
κ2 A h(1) − h(2) e[h −2]γ0 +2γ − h 2 j=2 hẋ, fj i
2
, (7.1436)
D−2
hẍ, fj i + [(1 + η0 )γ̇0 − γ̇] hẋ, fj i = 0 (j = 2, . . . , n). (7.1437)
γ = kγ0 , (7.1439)
where k is a constant.
By substituting the functions hẋ, fj i into equation (7.561) we obtain the following integrable by
quadrature ordinary differential equation
µ ¶
h(2)
γ̈0 − k − 1 + + ζ0 γ̇02 =
2
(7.1440)
" n
#
D − 2 2(k−1)γ0 2 ¡ (1) ¢ (1) h(2) X ¡ j ¢2 −2η0 γ0
e κ A h − h(2) eh γ0 − p e . (7.1441)
D−1 2 j=2
CBPF-MO-003/02 205
dt > 0, (7.1442)
The isotropic model. The isotropic model is described by the metric given by equation (7.530)
with
i
a ≡ ex = eγ0 /(D−1) , i = 1, . . . , n. (7.1443)
The scale factor a of the whole isotropically evolving space M1 × . . . × Mn can be obtained by
integration of equation (7.565) with
pj = 0, j = 2, . . . , n. (7.1444)
It can be proven that the above mentioned consistency condition leads to the following constraint
At first we present the special solution to equation (7.565). The special solution describes the
asymptotical behaviour of all solutions at late time. It is the steady state solution
"s #
2Λ(1 + ε0 )
a ∼ exp ts , ρ(1) = A ≡ Λ/κ2 , (7.1446)
(D − 1)(D − 2)
for h(1) 6= 2, where ts is the synchronous time. Moreover, we present the deceleration parameter
(We note that the time t becomes synchronous if the perfect fluid component appears as Λ-term,
i.e. h(1) = 2). This yields that equation (7.565) under the condition given by equation (7.569) is
integrable by elementary methods. The result is
µ ¶1/[(D−1)(h(1) −h(2) −2ζ0 )]
τ2
a = a0 , (7.1452)
1 − τ2
(1)
ρ(1) = Aa(D−1)[h −2] , (7.1453)
D−1£ ¡ ¢ ¤
q = −1 + 2 − h(1) + h(1) − h(2) − 2ζ0 (1 − τ 2 ) , (7.1454)
2
(2) (1)
ρ /ρ = (1 + ε0 )/τ 2 − 1, (7.1455)
· ¸
(1) (2)
¡ (2)
¢ 1 − τ2 (1)
p + p̃ = (1 + ε0 ) 1 − h − 2ζ0 2
+1−h ρ(1) , (7.1456)
τ
Now we analyze the role of viscosity in this isotropic 2-component model using the obtained
exact solution. The main features of the model are the following. Under the weak energy condition
(ρ(1) + ρ(2) ≥ 0, ρ(1) + ρ(2) + p(1) + p̃(2) ≥ 0), which leads to the following restriction on the
parameters 2 ≥ h(1) > h(2) + 2ζ0 , the Universe expands eternally from the initial singularity. Near
2/[(2−h(2) −2ζ0 )(D−1)]
the singularity we obtain in the main order a ∼ ts There is only nonsingular solution
given by equation (7.571).
All solutions describes the accelerated expansion at least on the late phase of evolution if h(1) >
2(D − 2)/(D − 1). The solutions given by equation (7.577) describe the period of decelerated
expansion if h(2) + 2ζ0 < 2(D − 2)/(D − 1). We note that the cosmic deceleration phase is important
within mechanism of the clumping of matter into galaxies (see, for instance, [430]). Under the
assumption 2 ≥ h(1) > 2(D − 2)/(D − 1) > h(2) + 2ζ0 the decelerated expansion takes place during
the time interval (t0 , t∗ ), where t∗ is defined by
q (2) −2ζ
arcosh 1 + 2(D−2)/(D−1)−h
h(1) −2(D−2)/(D−1)
0
∗
t − t0 = r i. (7.1458)
A(D−1) ¡ D−2
¢h 2(D−2)/(D−1)−h (2) −2ζ
κ2 2(D−2) (h(1) − h(2) ) h(1) − 2 D−1 1 + h(1) −2(D−2)/(D−1) 0
Equation (7.583) shows that introducing of the bulk viscosity reduces the phase of the decelerated
expansion.
The density ratio given by equations (7.574) and (7.1455) exhibits the following property
2ζ0
limt→+∞ ρ(2) /ρ(1) = ε0 ≡ . (7.1459)
h(1) − h(2) − 2ζ0
So, the bulk viscosity allow to resolve the coincidence problem which appears in the unviscous
2-component model.
Now we consider the general anisotropic behaviour of the model. Once integrating equation
(7.565) we get the first integral of the form γ̇0 = F (γ0 , C), where F is some function and C is an
arbitrary constant. Substituting γ̇0 and the functions hẋ, fj i given by equation (7.563) into equations
(7.552) and (7.554) we express ρ(2) and nT ṡ via γ0 . The subsequent analysis of the expressions for
γ0 , ρ(2) and nT ṡ shows the presence of solutions with physically nonsatisfactory behaviour near the
initial singularity. To exclude such solutions we put the following restrictions on the parameters
These restrictions guarantee the following properties for all solutions: the density ρ(2) and the
variation rate of entropy ṡ are positive during the evolution, which starts at the initial singularity
of the Kasner type and proceeds eternally to the subsequent isotropic expansion; the density ratio
ρ(2) /ρ(1) tends to the nonzero constant as t → +∞ (see equation (7.584)).
To study these behaviour in detail let us obtain the exact solution. We note that the equation
γ̇0 = F (γ0 , C) is integrable by quadrature for arbitrary parameters h(1) , h(2) , ζ0 and η0 . To express
the exact solution in elementary functions we put the following relation on the parameters
For the nonviscous model (η0 = ζ0 = 0) the relation reads h(1) = 2h(2) . The latter corresponds, for
instance, to the so called ΛCDM cosmological model with Λ-term (h(1) = 2) and dust (h(2) = 1).
We remind that the parameters obey the inequalities given by formula (7.585). Comparing equation
(7.586) and formula (7.585), one gets h(2) > 0.
Now we start the integration procedure in the time gauge defined by equations (7.564) and
(7.576) . The integration of equation (7.565) gives
τ (sin α + τ cos α)
eβγ0 (t) = C0 , (7.1462)
1 − τ2
Pn j 2
where we cancelled the constant j=2 (p ) by introducing the following constants
v q Pn
u 2h(2) j 2
j=2 (p )
n
X
u 2h(2) κ2 A(h(1) −h(2) )
C0 ≥ t (pj )2 , α = arcsin ∈ [0, π/2], (7.1463)
κ2 A(h(1) − h(2) ) j=2
C0
The variable τ was introduced by equation (7.582). Substituting equation (7.587) into equation
(7.563) and taking into account equations (7.564) and (7.576), we obtain by integration
s · ¸1/β
(D − 2)β j τ
hx, fj i = P p ln + qj , (7.1465)
(D − 1)h(2) nj=2 (pj )2 sin α + τ cos α
where q j are arbitrary constants. Substituting equations (7.587) and (7.590) into the decomposition
given by equation (7.559), we get
· ¸1/β · ¸1/β
(sin α + τ cos α)2 u τ
x = ln C0 + ln r + s, (7.1466)
1 − τ2 hu, ui τ (sin α + τ cos α)
CBPF-MO-003/02 208
Owing to the orthogonality property given by equation (7.558) the coordinates ri and si of these
vectors in the canonical basis e1 , . . . , en satisfy the following constraints
n
X n
X X n
i
¡ i ¢2 (D − 2)β 1
di r = hu, ri = 1, di r = hr, ri − di dj ri rj = (2)
+ , (7.1468)
i=1 i=1 i,j=1
(D − 1)h D − 1
n
X
di si = hu, si = 0. (7.1469)
i=1
The constants ri may be called Kasner-like parameters, because of the existence of the constraints
given by equation (7.593). We remind, that the coordinates of the vector x in the canonical basis
are the logarithms of the scale factors ai ≡ exp[xi ].
Finally, we present the exact solution:
· ¸(2−h(2) )/β · ¸2/(D−1)β
2 τ (sin α + τ cos α) 2 (sin α + τ cos α)2
ds = − C0 dt + C0 ×
1 − τ2 1 − τ2
(7.1470)
Xn · ¸ 2r i /β
τ i
× e2s ds2i , (7.1471)
i=1
sin α + τ cos α
· ¸−(2−h(2) )/β
(1) τ (sin α + τ cos α)
ρ = A C0 , (7.1472)
1 − τ2
· ¸
(2) (1) β 2
ρ /ρ = (1 + ε0 )F12 (τ ) 1 − (2) F2 (τ ) − 1, (7.1473)
h
r · ¸
D−1 η0 β 2
nT ṡ = 8κ2 A (1 + εo ) F1 (τ ) ζ0 + (2) F2 (τ ) ρ(1) ,
3 3
(7.1474)
D−2 h
where
1
2
(1+ τ 2 ) sin α + τ cos α (1 − τ 2 ) sin α
F1 (τ ) = , F2 (τ ) = . (7.1475)
τ (sin α + τ cos α) (1 + τ 2 ) sin α + 2τ cos α
The solution has the following integration constants: C0 > 0, α ∈ [0, π/2], t0 , r1 , . . . , rn , s1 , . . . , sn .
The constants ri and si satisfy the constraints given by equations (7.593) and (7.1469). Then the
number of free integration constants is 2n as required. The limit for α → +0 of this exact solution
is the isotropic solution obtained in section 3.2.
Before we start the studying of the obtained exact solution near the initial singularity let us
remind the multi-dimensional generalization of the well-known Kasner solution [61]. It reads (for
the synchronous time ts ) as follows
n
X i
ds2 = −dt2s + Ai t2ε 2
s dsi . (7.1476)
i=1
CBPF-MO-003/02 209
Such a metric describes the evolution of a vacuum model under consideration. The Kasner param-
eters εi satisfy the constraints
n
X n
X
i
¡ ¢2
di ε = 1, di εi = 1. (7.1477)
i=1 i=1
The generalized Kasner solution describes the contraction of some spaces from the set M1 , . . . , Mn
and the expansion for the other ones. According to equation (7.602), the number of either contract-
ing or expanding spaces depends on n (the total number of spaces) and di=1,n (their dimensions).
We note that the constraints given by equations (7.593) for ri coincide with these constraints
for εi when ζ0 = η0 = 0 and h(1) = 2h(2) , i.e. the Kasner-like parameters ri become exactly Kasner
parameters εi in the absence of viscosity. Therefore, if the parameters ζ0 and η0 are small enough,
then the model describes a behaviour of the Kasner type as τ → +0, i.e. towards to the initial
singularity. However, too strong viscosity suppress a behaviour of the Kasner type. It can be shown
that if the parameters ζ0 = η0 = 0 are large enough, then the model describes expansion of all factor
spaces M1 , . . . , Mn near the initial singularity.
One can prove that the final stage of the evolution (τ → 1 − 0) exhibits the isotropic expansion.
The asymptotical behaviour of the model for τ → 1 − 0 is described by the exact solution given by
equations (7.571)-(7.575).
CBPF-MO-003/02 210
Here we consider a cosmological model from Chapter 7 with the metric (7.876) on the manifold
(7.877) describing the ”evolution” of n Einstein spaces in the presence of (m+1)-component perfect-
fluid matter with pressures in all spaces proportional to the density
(α)
(α) ui
pi (t) = (1 − )ρ(α) (t), (8.1478)
di
(α)
where ui = const, i = 1, . . . , n; α = 0, . . . , m.
In the previous section the model was reduced to the Lagrangian system with the Lagrangian
(7.899), where
n m
1X i X (α)
V = V (x) = − ξ di exp(−2xi + 2γ0 (x)) + κ2 A(α) exp(ui xi ). (8.1479)
2 i=1 α=0
is the potential. The relation (8.1479) may be also presented in the form
m̄
X (α)
V = Aα exp(ui xi ), (8.1480)
α=0
In this case the 00-component of eq. (8.1483) is satisfied and the set (ea , a = 1, . . . , n − 1) is defined
up to O(n − 1)-transformation.
CBPF-MO-003/02 212
Remark. A special example of the diagonalization with the relations (8.1486) and
Xn n
X n
X
a
z = eai xi = [da /( dj )( dj )]1/2
dj (xj − xi ), (8.1487)
j=a j=a+1 j=a+1
For Λ-term and curvature components (see (8.1482) and (8.1481)) we have
j = 1, . . . , n. The calculation of
(see (8.1486)) we have in this limit v → 0. Under certain additional assumptions the limit (8.1495)
describes the approaching to the singularity. We impose the following restrictions on the parameters
uα in the potential (8.1489) for components with Aα 6= 0:
We note that due to (8.1492) the second condition is always satisfied for Λ-term and curvature
components (i.e. for α = 0, m + 1, . . . , m + n = m̄).
We restrict the Lagrange system (8.1488) on V− , i.e. we consider the Lagrangian
L− ≡ L|T M− , M− = V− × R+ , (8.1499)
where T M− is tangent vector bundle over M− and R+ ≡ {N > 0}. (Here F |A means the restriction
of function F on A.) Introducing an analogue of the Misner-Chitre coordinates in V− [278, 279]
1 + ~y 2
z 0 = − exp(−y 0 ) , (8.1500)
1 − ~y 2
~y
~z = −2 exp(−y 0 ) , (8.1501)
1 − ~y 2
|~y | < 1, we get for the Lagrangian (8.1488)
1 0
L− = N −1 e−2y [−(ẏ 0 )2 + hij (~y )ẏ i ẏ j ] − N V. (8.1502)
2
Here
hij (~y ) = 4δij (1 − ~y 2 )−2 , (8.1503)
i, j = 1, . . . , n − 1, and
m̄
X
V = V (y) = Aα exp Φ̄(y, uα ), (8.1504)
α=0
where
0
Φ̄(y, u) ≡ −e−y (1 − ~y 2 )−1 [u0 (1 + ~y 2 ) + 2~u~y ], (8.1505)
We note that the (n − 1)-dimensional open unit disk (ball)
with the metric h = hij (~y )dy i ⊗ dy j is one of the realization of the (n − 1)-dimensional Lobachevsky
space H n−1 .
We fix the gauge
N = exp(−2y 0 ) = −z 2 . (8.1507)
Then, it is not difficult to verify that the Lagrange equations for the Lagrangian (8.1502) with the
gauge fixing (8.1507) are equivalent to the Lagrange equations for the Lagrangian
1 1
L∗ = − (ẏ 0 )2 + hij (~y )ẏ i ẏ j − V∗ (8.1508)
2 2
with the energy constraint imposed
1 1
E∗ = − (ẏ 0 )2 + hij (~y )ẏ i ẏ j + V∗ = 0. (8.1509)
2 2
CBPF-MO-003/02 214
Here
m̄
X
−2y 0
V∗ = e V = Aα exp(Φ(y, uα )), (8.1510)
α=0
where
Φ(y, u) = −2y 0 + Φ̄(y, u). (8.1511)
Now we are interested in the behavior of the dynamical system in the limit y 0 → −∞ (or,
equivalently, in the limit z 2 = −(z 0 )2 + (~z)2 → −∞, z 0 < 0) implying (8.1495). Using the relations
(u0 6= 0 )
A(~y , −~u/u0 )
Φ(y, u) = −u0 exp(−y 0 ) − 2y 0 , (8.1512)
1 − ~y 2
A(~y , ~v ) ≡ (~y − ~v )2 − ~v 2 + 1, (8.1513)
we get
lim exp Φ(y, u) = 0 (8.1514)
y 0 →−∞
θ∞ (x) ≡ + ∞, x ≥ 0,
0, x < 0. (8.1516)
Using restrictions (8.1497), (8.1498) and relations (8.1510), (8.1514), (8.1515) we obtain
X
V∞ (~y ) ≡ 0lim V∗ (y 0 , ~y ) = θ∞ (−A(~y , −u~α /uα0 )). (8.1517)
y →−∞
α∈∆+
Here we denote
∆+ ≡ {α|(uα )2 > 0}. (8.1518)
We note that due to (8.1494) Λ-term and curvature components do not contribute to V∞ (i.e. they
may be neglected in the vicinity of the singularity).
The potential V∞ may be also written as following
V∞ (~y ) = V (~y , B) ≡ 0, ~y ∈ B,
+∞, ~y ∈ Dn−1 \ B, (8.1519)
where \
B= B(~v α ) ⊂ Dn−1 , (8.1520)
α∈∆+
p
B(~v α ) = {~y ∈ Dn−1 ||~y − ~v α | > (~v α )2 − 1}, (8.1521)
and
~v α = −~uα /uα0 , (8.1522)
α ∈ ∆+ .
CBPF-MO-003/02 215
y 0 = ω(t − t0 ), (8.1526)
flow on the Lobachevsky space H n−1 = (Dn−1 , hij dy i ⊗ dy j ). In this case there are two families of
non-trivial geodesic solutions (i.e. y(t) 6= const):
√ √
1. ~y (t) = ~n1 [ v 2 − 1 cos ϕ(t) − v] + ~n2 v 2 − 1 sin ϕ(t), (8.1529)
√ 1
ϕ(t̄) = 2 arctan[(v − v 2 − 1) tanh( 2 ω(t − t1 )], (8.1530)
2. ~y (t) = ~n2 tanh( 21 ω(t − t1 )). (8.1531)
Here ~n21 = ~n22 = 1, ~n1~n2 = 0, v > 1, ω > 0, t1 = const.
Graphically√the first solution corresponds to the arc of the circle with the center at point (−v~n1 )
and the radius v 2 − 1. This circle belongs to the plane spanned by vectors ~n1 and ~n2 (the centers of
the circle and the ball Dn−1 also belong to this plane). We note, that the solution (8.1529)-(8.1530)
in the limit v → ∞ coincides with the solution (8.1531).
We note, that the boundary of the billiard ∂B is formed by geodesics. For some billiards this
fact may be used for ”gluing” certain parts of boundaries.
When ∆+ 6= ∅ the Lagrangian (8.1527) describes the motion of the particle of unit mass, moving
in the (n − 1)-dimensional billiard B ⊂ Dn−1 (see (8.1520)). The geodesic motion in B (8.1529)-
(8.1531) corresponds to a ”Kasner epoch” and the reflection from the boundary corresponds to the
change of Kasner epochs. For n = 3 some examples of (2-dimensional) billiards are depicted in Figs.
2-4.
FIGURE 4
The billiard B in Fig. 2. has an infinite volume: volB = +∞. In this case there are three
open zones at the infinite circle |~y | = 1. After a finite number of reflections from the boundary the
particle moves toward one of these open zones. For corresponding cosmological model we get the
”Kasner-like” behavior in the limit t → −∞ [61] (see subsection 8.1.3.3 below).
For billiards depicted in Figs. 3 and 4 we have volB < +∞. In the first case (Fig. 3) the
closure of the billiard B̄ is compact (in the topology of Dn−1 ) and in the second case (Fig. 4) B̄ is
non-compact. In these two cases the motion of the particle is (probably) stochastic.
Analogous arguments may be applied to the case n > 3. So, we are interested in the configura-
tions with finite volume of B.
We propose a simple geometric criterion for the finiteness of the volume of B and compactness
of B̄ in terms of the positions of the points (8.1522) with respect to the (n − 2)-dimensional unit
sphere S n−2 (n ≥ 3). We say that the point ~y ∈ S n−2 is (geometrically) p illuminated by the point-like
2
source located at the point ~v , |~v | > 1, if and only if |~y − ~v | ≤ |~v | − 1. In Fig. 1 the source P
illuminates the closed arc [P1 , P2 ]. We also say that the point ~y ∈ S n−2 is strongly p illuminated by
the point-like source located at the point ~v , |~v | > 1, if and only if |~y − ~v | < |~v |2 − 1. In Fig. 1
the source P strongly illuminates the open arc (P1 , P2 ). The subset N ⊂ S n−2 is called (strongly)
illuminated by point-like sources at {~v α , α ∈ ∆+ } if and only if any point from N is (strongly)
illuminated by some source at ~v α (α ∈ ∆+ ).
Proposition 8.1.1. The billiard B (3.26) has a finite volume if and only if the point-like sources
of light located at the points ~v α (8.1522) illuminate the unit sphere S n−2 . The closure of the billiard
B̄ is compact (in the topology of Dn−1 ' H n−1 ) if and only if the sources at points (8.1522) strongly
illuminate S n−2 .
Proof. We consider the set ∂ c B ≡ B c \ B̄, where B c is the completion of B (or, equivalently,
the closure of B in the topology of Rn−1 ). We remind that B̄ is the closure of B in the topology
of Dn−1 . Clearly, that ∂ c B is a closed subset of S n−2 , consisting of all those points that are not
strongly illuminated by sources (8.1522). There are three possibilities: i) ∂ c B is empty; ii) ∂ c B
contains some interior point (i.e. the point belonging to ∂ c B with some open neighborhood); iii)
∂ c B is non-empty finite set, i.e. ∂ c B = {~y1 , . . . ~yl }. The first case i) takes place if and only if B̄ is
CBPF-MO-003/02 217
compact in the topology of Dn−1 . Only in this case the sphere S n−2 is strongly illuminated by the
sources (8.1522). Thus the second part of proposition is proved. In the case i) volB is finite. For
the volume we have Z Z 1
n−1
√
volB = d ~y h = dr(1 − r2 )1−n Sr . (8.1532)
B 0
The ”area” Sr → C > 0 as r → 1 in the case ii) and, hence, the integral (8.1532) is divergent. In
the case iii)
Sr ∼ C1 (1 − r)2(n−2) as r → 1 (8.1533)
(C1 > 0) and, so, the integral (8.1532) is convergent. Indeed, in the case iii), when r → 1, the ”area”
Sr is the sum of l terms. Each of these terms is the (n − 2)-dimensional ”area” of a transverse side
of a deformed pyramid with a top at some point ~yk , k = 1, . . . , l. This multidimensional pyramid is
formed by certain parts of spheres orthogonal to S n−2 in the point of their intersection ~yk . Hence,
all lengths of the transverse section r = const of the ”pyramid” behaves like (1 − r)2 , when r → 1,
that justifies (8.1533). But the unit sphere S n−2 is illuminated by the sources (8.1522) only in the
cases i) and iii). This completes the proof.
The problem of illumination of convex body in multidimensional vector space by point-like
sources for the first time was considered in [274, 275]. For the case of S n−2 this problem is equivalent
CBPF-MO-003/02 218
to the problem of covering the spheres with spheres [276, 277]. There exist a topological bound on
the number of point-like sources m+ illuminating the sphere S n−2 [46]:
m+ ≥ n. (8.1534)
(ω > 0) instead of (8.1526). In the limit t → t0 + 0 we have y 0 → −∞ and ~y (t) → ~y0 ∈ B. So, the
stochastic behavior near the singularity is absent in this case.
8.1.3. Some examples and Kasner-like parametrization
8.1.3.1. Bianchi-IX cosmology.
Here we consider the well-known mixmaster model [269, 270] with the metric
3
X
g = ĝmix ≡ − exp[2γ(t)]dt ⊗ dt + exp[2xi (t)]ei ⊗ ei , (8.1537)
i=1
i.e. a triangle from Fig. 4 (see also [280]). In this case the circle S 1 is illuminated by sources at
points ~v i , i = 1, 2, 3, but not strongly illuminated. In agreement with Proposition 8.1.1 the billiard
B has finite volume, but B̄ is not compact.
8.1.3.2. Mixmaster cosmology with a chain of Einstein internal spaces
Now we consider the cosmological model with the metric
n
X
g = ĝmix + exp[2xi (t)]ĝ i , (8.1545)
i=4
CBPF-MO-003/02 220
The Λ-term and curvature components in the potential (8.1547) do not form walls near the sin-
gularity since the corresponding (u(α) )2 are negative (see (8.1494) and (8.1517)). For ”mixmaster”
components we have
(1) (2)
(ui ) = (4, 0, 0, 2d4 , . . . , 2dn ), (ui ) = (0, 4, 0, 2d4 , . . . , 2dn ),
(3) (4)
(ui ) = (0, 0, 4, 2d4 , . . . , 2dn ), (ui ) = (0, 2, 2, 2d4 , . . . , 2dn ),
(5) (6)
(ui ) = (2, 0, 2, 2d4 , . . . , 2dn ), (ui ) = (2, 2, 0, 2d4 , . . . , 2dn ). (8.1548)
The calculation of (uα )2 gives the same result as in the pure mixmaster case, see (8.1543).
P (k)
From (8.1548) we get ni=1 ui = 2(D − 2) > 0 and hence uk0 > 0, k = 1, . . . , 6 (see (8.1491)).
For curvature and Λ-term components uα0 > 0 due to (8.1492). So, the conditions (8.1497) and
(8.1498) are satisfied. Thus, we have m+ = 3 and the inequality (3.40) is not satisfied in our case
(n ≥ 4). There exists a non-empty ”shadow domain” in the sphere S n−2 . The billiard (8.1520) has
an infinite volume (Proposition 8.1.2). The oscilating behaviour near the singularity is absent for
the considered extension of the Bianchi-IX model.
8.1.3.3. Kasner-like parametrization.
In z-variables (8.1500), (8.1501) the geodesic solutions (8.1529)-(8.1531) with y 0 from (8.1496)
may written in the following form
where
1
th = −exp(−2ω(t − t0 )) < 0 (8.1550)
2ω
is a harmonic time variable, satisfying dth = eγ−γ0 dt. In (8.1549)
ω exp(ω(t0 − t1 ))
c= exp(ω(t1 − t0 )), c∗ = − , (8.1551)
cos ϕ0 2 cos ϕ0
√
wheere ϕ0 = 2 arctan(v − v 2 − 1), sin ϕ0 = v1 for the circle (8.1529) (0 < ϕ0 < π2 ) and ϕ0 = 0 for
the line (8.1531) and vectors
xi = αi ln ts + β i , (8.1555)
where
alphai = (ei0 + eis ns )/q = eia na /q, β i = eis (ns∗ − ns ) + αi ln(qc) (8.1556)
i = 1, . . . , n, s = 1, . . . , n − 1 and (na ) = (1, ~n). It follows from (8.1484) and (8.1486) that
where (na ) = (1, ~n) and set (αi ) is defined by (8.1556). It follows from (8.1562) that the point
~n ∈ S n−2 belongs to the domain of shadow, when the ”Kasner sphere” S n−2 is illuminated by the
sources (8.1522), if and only if
Xn
(ν)
ui αi > 0, ν ∈ ∆+ . (8.1563)
i=1
The Kasner-like parameters in (8.1559), (8.1560) should satisfy the inequalities (8.1563). The
criteria of the finiteness of the billiard volume (see Proposition 8.1.1) may be reformulated in terms
of inequalities on the Kasner-like parameters:
Proposition 8.1.3. The billiard B (8.1520) has a finite volume if and only if the set of in-
equalities (8.1563) is inconsistent.
If there exist solutions of (8.1563) then the billiard B has infinite volume and there are open
zones on Kasner sphere. In this case we have Kasner-like asymptotes (8.1559), (8.1560), (8.1563).
Example: prototype model. Let us consider the multicomponent perfect fluid cosmology
with d1 = . . . = dn = 1 and the potential
1 X
V = exp[2γ0 (x) + 2(xi − xj − xk )] (8.1564)
4
(i,j,k)∈S
in the Lagrangian (7.899) (and minisupermetric Gij = δij − 1), where S = {(i, j, k)|i, j, k =
1, . . . , n; j 6= i 6= k, j < k}, n > 2. This model may be considered as a prototype cosmolog-
ical model describing the behaviour of the solutions to the Einstein equations in the dimension
D = 1 + n near the singularity [32, 80]. The sum in (8.1564) contains n(n − 1)(n − 2)/2 terms.
For any component α ∈ S the trivial calculation gives (u(α) )2 = 8 > 0 and uα0 > 0 (see (7.911) and
(8.1491)). Thus the restrictions (8.1497) and (8.1498) are satisfied. The corresponding billiard has
a finite volume for n < 10 and infinite volume for n ≥ 10. This was proved in fact in [80] using the
inequalities on Kasner parameters. This billiard is not compact for all dimensions, since n points
on the Kasner sphere corresponding to the sets (αi ) = (1, 0, . . . , 0), . . . , (0, 0, . . . , 1) are not strongly
illuminated. From (8.1563) and (8.1564) we get the inequalities on Kasner parameters
n
X
αl + αi − αj − αk = 1 + αi − αj − αk > 0, (8.1565)
l=1
Here, like in the Bianchi-IX case [278, 279], the considered reduction scheme uses a special
time gauge (or parametrization of time). As it was pointed in [280] one should be careful in the
interpretations of the results of computer experiments for other choices of time.
8.1.4.1. More general restrictions on parameters.
Here we discuss the physical sense of the restrictions on parameters of the model (8.1497) and
(8.1498). The condition (8.1497) means that the densities of the ”anisotropic” components with
(u(α) )2 > 0 should be positive. Using (8.1478) and (8.1491) we rewrite the restriction (8.1498) in
the equivalent form
X n (α)
ρ(α) − pi
di > 0, (8.1566)
i=1
ρ(α)
(We remind that, h is the metric on Lobachevsky space Dn−1 .) The form of WDW eq. (8.1571)
follows from the demands of minisuperspace invariance and conformal covariance [66, 67, 55]. Using
relations µ ¶2
∂
∆[Ḡ] = − + ∆[h], R[Ḡ] = R[h] = −(n − 1)(n − 2), (8.1573)
∂y 0
we rewrite (8.1571) in the form
à µ ¶2 !
1 ∂ 1 1
− ∆[h] − (n − 2)2 + V∗ Ψ = 0. (8.1574)
2 ∂y 0 2 8
where ∂B = B̄ \ B is the boundary of the billiard B from (8.1520) (in Dn−1 ) and
1
∆∗ [h] = ∆[h] + (n − 2)2 . (8.1576)
4
Now, we suppose that B̄ is compact and the operator (8.1576) with the boundary condition (8.1575)
has a negative spectrum, i.e.
∆∗ [h]Ψn = −En2 Ψn , (8.1577)
En > 0, n = 0, 1, . . . (this is valid at least for ”small enough” B). Using (8.1577) we get the general
solution of the asymptotic WDW eq. (8.1575)
∞
X
0
Ψ∞ (y , ~y ) = [cn exp(−iEn y 0 )Ψn (~y ) + c∗n exp(iEn y 0 )Ψ∗n (~y )], (8.1578)
n=0
that may be considered as a starting point for the construction of third quantized models in the
vicinity of the singularity [282]. We also note the considered here scheme is also applicable for the
case n = 2 .
CBPF-MO-003/02 225
z a = eaA xA , (8.1580)
diagonalizing the minisuperspace metric, i.e. Ḡ = ηab dz a ⊗dz b where (ηab ) = (η ab ) ≡ diag(−1, +1, . . . , +1),
and here and in what follows a, b = 0, . . . , N − 1; N = n + l. The matrix of linear transformation
(eaA ) satisfies the relation
ηab eaA ebB = ĜAB (8.1581)
or, equivalently,
η ab = eaA ḠAB ebB = (ea , eb ), (8.1582)
where ea = (eaA ).
Inverting the map (8.1580) we get
xA = eA a
az , (8.1583)
where for components of the inverse matrix (eA a −1
a ) = (eA ) we obtain from (8.1582)
eA
a = Ĝ
AB b
eB ηba . (8.1584)
and hence n
X
z 0 = e0A xA = q −1 di xi . (8.1586)
i=1
0
In z-coordinates (8.1580) with z from (8.1586) the Lagrangian (8.1579) reads
1
LQ = LQ (z a , ż a , N ) = N −1 ηab ż a ż b − N V (z), (8.1587)
2
CBPF-MO-003/02 226
where X
V (z) = Ar exp(2Uar z a ) (8.1588)
r∈S∗
is a potential,
S∗ = {Λ} ∪ {1, . . . , n} ∪ S (8.1589)
is an index set and
w 1
AΛ = −wΛ, Aj = ξj dj , As = εs Q2s , (8.1590)
2 2
j = 1, . . . , n; s ∈ S. Here we denote
Uar = eA r r b
a UA = (U , e )ηba , (8.1591)
a = 0, . . . , N − 1; r ∈ S∗ (see (8.1584)). Remind that product (U, U 0 ) was defined in (3.82) and U r
in (2.40), (??), (4.420).
From (3.82)-(3.84), (8.1585) and (8.1591) we deduce
n
X
U0r r 0
= −(U , e ) = ( Uir )/q(D − 2), (8.1592)
i=1
r ∈ S∗ . For the Λ-term and curvature-term components we obtain from (8.1585) and (8.1592)
j = 1, . . . , n.
For ”brane” components we get ¿from (2.40), (8.1585) and (8.1592)
p
U0s = d(Is )/ (D − 2)(D − 1) > 0. (8.1594)
s ∈ S. For θa = 1, a ∈ ∆, and ε[g] = −1, the first restriction means that all ε(Is ) = 1, s ∈ S, i.e.
all p-branes are either Euclidean or contain even number of “times”. Restriction (ii) implies
µ ¶
s s d(Is )
(U , U ) = d(Is ) 1 + + λ2as > 0, (8.1598)
2−D
where λ2 = λα λβ hαβ , D > 2. As we shall see below, both restrictions are necessary for a formation
of potential walls in the Lobachevsky space when a certain asymptotic in time variable is considered.
CBPF-MO-003/02 227
Here we consider a behaviour of the dynamical system, described by the Lagrangian (8.1587)
with the potential (8.1588) for N ≥ 3 in the limit z 0 → −∞, z = (z 0 , ~z) ∈ V− , where V− ≡
{(z 0 , ~z) ∈ RN |z 0 < −|~z|} is the lower light cone.
Due to relations (8.1590), (8.1593)-(8.1595), (8.1596) and (8.1598) the parameters U r in the
potential (8.1588) obey the following restrictions:
~ r )2 > 0;
1. Ar > 0 for (U r )2 = −(U0r )2 + (U (8.1599)
2. U0r > 0 for all r ∈ S∗ . (8.1600)
According the results of subsection 8.1 the cosmological model under consideration with the
restrictions 1. and 2. imposed in the limit z 0 → −∞, z ∈ V− , is reduced the motion of a point
particle in a billiard belonging to Lobachevsky space. In our case this billiard reads
\
B= Bs ⊂ DN −1 , (8.1601)
s∈S
m ≥ N = n + l. (8.1605)
The latter is a restriction the number of p-branes m = |S| that allow the existence of oscillating
(e.g. stochastic) behaviour near the singular point.
Now, let us consider Kasner-like solutions
n
X i
g = wdτ ⊗ dτ + Ai τ 2α ĝ i , (8.1606)
i=1
β
ϕβ = αβ ln τ + ϕ0 , (8.1607)
Xn Xn
i
di α = di (αi )2 + αβ αγ hβγ = 1, (8.1608)
i=1 i=1
a
F =0 (8.1609)
s ∈ S, then the field configuration (8.1606)-(8.1609) is the asymptotical (attractor) solution for a
family of (exact) solutions with non-zero charges: Qs 6= 0, when τ → +0.
Relations (8.1610) may be easily understood using relations (3.217). Indeed, from (3.217) and
zero value limits for forms F a , a ∈ ∆, we get
s (α)
exp[2U s (x)] = Cs τ 2U → 0, Cs 6= 0, (8.1611)
αA = eA a
a n /q, (na ) = (1, ~n), ~n ∈ S N −2 . (8.1612)
Here we use the diagonalizing matrix (eA a ) and the parameter q defined in subsection 8.2.1 (see
(8.1585)).
Proposition 8.2.1. Let us consider a point-like source located at a point ~v s from (8.1603).
A point ~n ∈ S N −2 is illuminated by this source if and only if the corresponding Kasner vector
α = (αA ) ∈ K defined by (8.1612) satisfies the relation
X
U s (α) = UAs αA = di αi − χs λas γ αγ ≤ 0. (8.1613)
i∈Is
~ sU
U ~ s0
s s0
~v ~v = s s0 = (8.1616)
U0 U0
· ¸
(D − 2)(D − 1) d(Is )d(Is0 ) αβ
=1+ d(Is ∩ Is0 ) + + χs χs0 λaα λbβ h . (8.1617)
d(Is )d(Is0 ) 2−D
CBPF-MO-003/02 229
Here a billiard representation was considered when the restrictions (8.1596) and (8.1597) are
imposed. Now we relax the first restriction, i.e. we put
εs = −1, (8.1618)
for some s ∈ S. Relation (8.1618) occurs when spherically symmetric solutions with p-branes are
considered. In this case we may obtain ”waterfall potentials” with V = −∞ instead of V = +∞
inside of “walls”. The “waterfall potentials” prevent the oscillating behaviour near the singularity
but meanwhile do not forbid the existence of solutions with Kasner-like asymptotical behaviour (if
there are open shadow zones). Let us consider the following example:
1 ∈ Is , d1 = 1, (8.1619)
for all s ∈ S, i.e. all “branes” overlap the one-dimensional space M1 . In this case the Kasner set
α1 = 1, αi = αβ = 0, (8.1620)
M = (u− , u+ ) × M1 × M2 (8.1621)
i.e. we have two electric branes corresponding to the form F 1 , and two magnetic branes correspond-
ing to F 2 . Branes s1 and s3 ”live” in M1 and branes s2 and s4 ”live” in M2 .
We put λ1 = λ2 = λ and λ2 = d/2.
Then from (8.1616) we get
i = 1, 2, 3, 4, and
~v s1 ~v s2 = ~v s1 ~v s3 = ~v s2 ~v s4 = ~v s3 ~v s4 = 0. (8.1625)
This means that the points ~v si , i = 1, 2, 3, 4, form a square in R2 containing S 1 (~v s1 = −~v s4 ,
~v s2 = −~v s3 ), i.e. all points of S 1 are illuminated by these four points. The billiard B (8.1601) is a
CBPF-MO-003/02 230
Figure 5. Square billiard in the 5-dimensional model with two internal spaces, scalar field and four
“branes” (two electric and two magnetic).
M = (u− , u+ ) × M1 × M2 × M3 (8.1626)
i.e. we have three electric branes corresponding to the form F 1 . The brane si ”lives” in Mi ,
i = 1, 2, 3.
From (8.1616) we get
i 6= j; i, j = 1, 2, 3.
For d ≥ 2 the points ~v si , i = 1, 2, 3, form a triangle in R2 containing S 1 and all points of
S 1 are illuminated by these three points. The billiard (8.1601) is a sub-compact triangle in the
Lobachevsky space D2 . For d = 2 it is depicted on Fig. 6.
For d = 1 we obtain (at least formally) the billiard B depicted on Fig. 7. The closure of B is
not compact but the area of B is finite. This billiard appears in the well-known Bianchi-IX model
For d = 1 the Restriction 1 on composite p-branes (from Chapter 2) is not satisfied but to avoid
CBPF-MO-003/02 231
Figure 6. Triangle billiard in the 7-dimensional model with three internal spaces and three electric
“branes”.
this obstacle we may consider non-composite case when the Lagrangian (8.1627) is replaced by the
following Lagrangian in the dimension D = 4 with three 2-forms F a , a = 1, 2, 3,
X3
1 a 2
L = R[g] − 2Λ − (F )g , (8.1630)
a=1
2!
8.2.4 D = 11 supergravity
8.2.4.1. ”Truncated” D = 11 supergravity
Now we consider a ”truncated” bosonic sector of D = 11 supergravity governed by the action
(3.109). In this case we have electric 2-branes (d(Is ) = 3) and magnetic 5-branes (d(Is ) = 6).
From (8.1616) we get
and
0
~v s~v s = = 1 + 10[d(Is ∩ Is0 ) − 1], vs = vs0 = e, (8.1634)
5
= 1 + [d(Is ∩ Is0 ) − 4], vs = vs0 = m, (8.1635)
2
= 1 + 5[d(Is ∩ Is0 ) − 2], vs = e, vs0 = m. (8.1636)
Scalar products (8.1632)-(8.1636) are (in some sense) “building blocks” for constructing billiards
for the model under consideration.
CBPF-MO-003/02 232
Now we suggest an example of a billiard with a finite volume that occurs in the truncated D = 11
supergravity. Let us consider the metric (7.1340) defined on the manifold
M = (u− , u+ ) × M1 × . . . × M5 , (8.1637)
where all (Mi , g i ), i = 1, . . . , 5, are 2-dimensional Einstein manifolds of the Euclidean signature and
w = −1.
We consider ten magnetic 5-branes wrapped on six-dimensional “submanifolds” Mi × Mj × Mk ,
1 ≤ i < j < k ≤ 5. Thus, the 5-branes are labeled by indices
1 ≤ i < j < k ≤ 5. √
It follows from (8.1633) and (8.1635) that all vectors ~v s have the same length: |~v s | = 6 and
0
~v s~v s = 1, −4 for dimensions of intersections d(Is ∩ Is0 ) = 4, 2 respectively.
Now we prove that in our case the set of sources located at points ~v s ∈ R4 , s ∈ S = Sm
”illuminates” the 3-dimensional (Kasner) sphere S 3 and hence (according to Proposition 8.1.1) the
4-dimensional billiard (8.1601) B ⊂ D4 has a finite volume. Due to Proposition 8.2.2 it is sufficient
to prove the following proposition.
Proposition 8.2.3. There are no α = (α1 , . . . , α5 ) ∈ R5 satisfying the relations
5
X X5
i 1
α = (αi )2 = , (8.1639)
i=1 i=1
2
and
αi + αj + αk > 0, (8.1640)
for all 1 ≤ i < j < k ≤ 5.
This proposition is a consequence of the following statement.
CBPF-MO-003/02 233
Proof. Let us consider the set K of Kasner vector parameters α = (α1 , . . . , α5 ) ∈ R5 satisfying
(8.1639). Let
G = {α ∈ K|α1 < α2 < α3 < α4 < α5 } (8.1643)
and Ḡ = {α ∈ K|α1 ≤ α2 ≤ α3 ≤ α4 ≤ α5 }. G is an open submanifold of the 3-dimensional
“Kasner” manifold K and Ḡ is a closure of G. Ḡ is compact subset of K. Let us consider a smooth
function f : R5 → R defined by the relation
f (α) = α1 + α2 + α3 . (8.1644)
αmax ∈ Ḡ. It is clear that C ≥ 0, since C ≥ f (α1 ) = 0, where α1 is defined in (8.1642). To prove
the proposition it is sufficient to prove that the point of maximum αmax is unique and
αmax = α1 . (8.1646)
Let us prove the relation (8.1646). The point αmax does not belong to G. Indeed, if we suppose
that αmax ∈ G we get the conditional extremum relation
dΦ(α) = 0 (8.1647)
at α = αmax , where
5
X 5
X
i
Φ(α) = Φ0 (α) ≡ f (α) + λ1 α + λ2 (αi )2 , (8.1648)
i=1 i=1
and λ1 and λ2 are Langrange multipliers. It follows from (8.1647) and (8.1648) that
∂i Φ = 1 + λ1 + 2λ2 αi = 0, i = 1, 2, 3, (8.1649)
∂j Φ = λ1 + 2λ2 αj = 0, j = 4, 5, (8.1650)
at α = αmax . Relations (8.1649) and (8.1650) imply λ2 6= 0 and α4 = α5 . The latter contradicts
the inequality α4 < α5 for points in G. Thus, αmax ∈ Ḡ \ G. The set ∂G = Ḡ \ G is a border of the
curved tetrahedron Ḡ. It is a union of four faces
six edges
α1 < α2 = α3 = α4 = α5 (V1 ),
α1 = α2 < α3 = α4 = α5 (V2 ),
α1 = α2 = α3 < α4 = α5 (V3 ),
α1 = α2 = α3 = α4 < α5 (V4 ).
The point of maximum αmax does not belong to any face Γa , a = 1, 2, 3, 4. Indeed, if we suppose
that αmax belongs to some face Γa we get the conditional extremum relation (8.1647) with modified
Φ:
Φ = Φ0 + λ3 (αa − αa+1 ), (8.1651)
a = 1, 2, 3, 4. But one may verify that there are no solutions of the relation (8.1647) in this case.
Analogous arguments lead us to a non-existence of points of maximum among the points of edges
E12 , E13 , E14 , E23 , E24 , E34 . Here the only difference is that there exist (only) two extremal points
belonging to E13 and E23 respectively:
µ ¶ µ ¶ µ ¶
1 2 1 12 3 4 1 3 5 1 18
α =α = 1− √ ,α = α = 1+ √ ,α = 1+ √ , (8.1652)
10 14 10 14 10 14
µ ¶ µ ¶ µ ¶
1 1 3√ 2 3 4 1 1√ 5 1 9√
α = 1− 6 ,α = α = α = 1− 6 ,α = 1+ 6 , (8.1653)
10 2 10 4 10 4
but f (α) < 0 in these points and hence they are not the points of maximum. Thus, αmax belongs
to the set of vertices:
1 1
α1 = − , α2 = α3 = α4 = α5 = (V1 ), (8.1654)
µ ¶ 2 4
1 2 1 3√ 3 4 5 1 ³ √ ´
α =α = 1− 6 , α =α =α = 1+ 6 (V2 ), (8.1655)
10 2 10
µ ¶
1 2 3 1 ³ √ ´ 4 5 1 3√
α =α =α = 1− 6 , α =α = 1+ 6 (V3 ), (8.1656)
10 10 2
1 7
α1 = α2 = α3 = α4 = − , α5 = (V4 ). (8.1657)
20 10
The calculation gives us f (α) < 0 for α = Vi , i = 2, 3, 4. Thus, the first vertex V1 = α1 is the point
of maximum. The Proposition 8.2.4 is proved.
We also proved the Proposition 8.2.3 (as a consequence of the Proposition 8.2.4) and due to
Proposition 8.2.2 the following proposition.
CBPF-MO-003/02 235
Proposition 8.2.5. For the model (8.1632)-(8.1638) the Kasner sphere S 3 is illuminated by the
set of ten sources located at points ~v s ∈ R4 from (8.1603) , with s = s(i, j, k), 1 ≤ i < j < k ≤ 5,
defined in (8.1638), and hence the billiard B (8.1601) has a finite volume.
Moreover, this proposition may be strengthen as follows.
Proposition 8.2.5a. In terms of Proposition 8.2.5 all points of the Kasner S 3 sphere except
five points ~n1 , . . . , ~n5 ∈ S 3 corresponding to the Kasner sets
µ ¶ µ ¶
1 1 1 1 1 1 1 1 1 1
α1 = − , , , , , . . . , α5 = , , , ,− ∈ K, (8.1658)
2 4 4 4 4 4 4 4 4 2
respectively (see (8.1612)) are strongly illuminated by the set of ten sources ~v s ∈ R4 , s ∈ S.
Proof. The set K of Kasner parameters satisfying (8.1639) is a union of 5! = 120 “sectors”
where
Ki1 i2 i3 i4 i5 = Ḡ = {α ∈ K|αi1 ≤ αi2 ≤ αi3 ≤ αi4 ≤ αi5 } (8.1660)
and (i1 , i2 , i3 , i4 , i5 ) is a permutation of (1, 2, 3, 4, 5). Due to Proposition 8.2.1a and Proposition 8.2.4
any point ~n = ~n(α) ∈ S 3 corresponding to α ∈ K12345 \ {α1 } is strongly illuminated by the source
located at point ~v s , with s = s(1, 2, 3). Analogously, any point ~n = ~n(α) ∈ S 3 corresponding to
α ∈ Ki1 i2 i3 i4 i5 \ {αi1 } is strongly illuminated by the source located at point ~v s , with s = s(i1 , i2 , i3 ),
where (i1 , i2 , i3 , i4 , i5 ) is a permutation of (1, 2, 3, 4, 5). The proposition is proved.
The points ~n1 , . . . , ~n5 from Proposition 8.2.5a are not strongly illuminated (see Proposition
8.2.1a). Thus, Proposition 8.2.5, Proposition 8.2.5a and Proposition 8.1.1 imply that the billiard
B (8.1601) has a finite volume but its closure B̄ is not compact. Points ~n1 , . . . , ~n5 ∈ S 3 are ending
points of five “horns” of B. (These “horns” look similar to potential energy “valleys” that occur in
some toy models, e.g. related to M(atrix) theory [283]).
Using relations (8.1612) one may verify that
1
~ni~nj = − , i 6= j, (8.1661)
4
i, j = 1, . . . , 5. This means that ~n1 , . . . , ~n5 are vertices of a 4-dimensional simplex.
Here we considered the billiard B generated by ten sources, corresponding to non-zero charges
: Qs 6= 0, s ∈ S. If some charges are zero: Qs = 0, s ∈ S0 , then the corresponding points ~v s , s ∈ S0
(S0 6= ∅) are “switched off” and billiard is generated by m ≤ 9 point-like sources. In this case we
have the following proposition.
Proposition 8.2.5b. In terms of Proposition 8.2.5 any subset of sources ~v s , s ∈ S \ S0 , with
S0 6= ∅, does not illuminate the Kasner sphere S 3 and hence the billiard B generated by this subset
has an infinite volume.
Proof. Without a loss of generality let us suppose that S0 contains s(1, 2, 3), i.e. at least the
source at ~v s , s = s(1, 2, 3), is ”switched off”. Then, the point ~n corresponding to the set α = (αi )
from (8.1656) belongs to the shadow, since
αi + αj + αk > 0, (8.1662)
for all 1 ≤ i < j < k ≤ 5, (i, j, k) 6= (1, 2, 3). Thus, the shadow set is non-empty. (This set is open
since it is an intersection of a finite number of open shadow sets corresponding to sources of light.)
The proposition is proved.
CBPF-MO-003/02 236
X 5
X
F = −e f
Qs τ (I¯s ), f =2 xi − γ, (8.1664)
s∈S i=1
where I¯s = {1, 2, 3, 4, 5} \ Is , s ∈ S = Sm , and S is the index set defined in (8.1638). For the
Chern-Simons term we obtain
X5
−2f
e F ∧F =2 ¯
Pi τ ({i}), (8.1665)
i=1
where X
2Pi = Qs Qs0 δ(Is ∩ Is0 , {i}), (8.1666)
s,s0 ∈S
(F ∧ F )(z) 6= 0, (8.1667)
in all points z ∈ M .
Proof. Let us suppose that the Chern-Simons term vanishes in some point (i.e. relation (8.1663)
¯ are linearly independent at any point, we obtain
in some point is satisfied). Since forms τ ({i})
Pi = 0, i = 1, 2, 3, 4, 5, or explicitly
where we denote Qijk = Qs for s = s(i, j, k), 1 ≤ i < j < k ≤ 5. Let us denote k2 = Q345 , k3 = Q245 ,
k4 = Q235 , k5 = Q234 and a = Q145 , b = Q135 . From (8.1672) we get
k3 k4
Q125 = − b − a. (8.1673)
k2 k2
From the relation k3 P3 + k4 P4 + k5 P5 = 0 we get
k5 k5
Q134 = − b − a. (8.1674)
k4 k3
CBPF-MO-003/02 237
k4 k5 k3 k5
Q123 = a, Q124 = b. (8.1675)
k2 k3 k2 k4
Substituting (8.1673)-(8.1675) into eq. (8.1668) we get
Hence a = b = 0. But this contradicts our supposition, that Qs 6= 0. So, the proposition is proved.
Thus, it follows from Proposition 8.2.5b and Proposition 8.2.6, that the inclusion of the Chern-
Simons term leads us to the billiard B of infinite volume. In this case some Kasner (shadow) zones
are opened and we have the Kasner-like behaviour near the singularity.
CBPF-MO-003/02 238
In this chapter we present several classes of solutions defined on the manifold (7.877) with block-
diagonal metric from (7.876). According to Lagrange representation ¿from the Chapter 4 (see also
Chapter 7) we deal with the exact solutions to the Lagrange equations for the following Lagrangian
1 1
L= exp(−γ + γ0 (x))Gij ẋi ẋj + ϕ̇2 − exp(γ − γ0 (x))V (x) (9.1680)
2 2
(compare with (7.899)), where V = Vc is the curvature part of potential, see (7.889).
Here ts > 0, ci , cϕ are non-zero constants and parameters αi , αϕ satisfy the Kasner-like relations
n
X
di αi = 1, (9.1684)
i=1
n
X
di αi2 + αϕ2 = 1. (9.1685)
i=1
This solution may be readily verified by substitution to Lagrange eguations for (9.1680) or by using
the relations for Ricci-tensor (7.884) and (7.884).
The special case of this solution with αϕ = 0, (i.e. in the absence of scalar field) was obtained
first in [61].
CBPF-MO-003/02 239
This solution in slyghtly diffent form was obtained also in [75, 76]. The solution is a special case
of the solution (7.1048)-(7.1053) with h = 1 and A = − 12 ξ1 d1 , since the ”1-curvature model” is
equivalent to a special case of the model with one-component perfect fluid ¿from (7.1045) with
h = 1.
Analogously, the power-law ”inflationary” solution for the negative curvature case ξ1 < 0 reads
(see (7.1055))
n
X
g = −dts ⊗ dts + A21 t2s ĝ 1 + A2i ĝ i , (9.1692)
i=2
ϕ = const, (9.1693)
where A21 = |ξ1 |/(d1 − 1) (see (7.1013), (7.1057)). We are lead here to the Milne-type solution from
[75].
There exist another parametrization of the solution (9.1686)- (9.1690) in terms of R-variable
related to τ -variable as
µ ¶d1 −1
R0
F = F (R) =1− = tanh2 (τ /2T ), ξ1 < 0, (9.1694)
R
µ ¶d1 −1
R0
= − 1 = tan2 (τ /2T ), ξ1 > 0. (9.1695)
R
p
Here R > R0 for ξ1 < 0 and R < R0 for ξ1 > 0; R0 = A1 21/(d1 −1) (d1 − 1)/|ξ1 |. In new variables
the metric and the scalar field may be written as
P
g = −F b−1 dR ⊗ dR + F b R2 A21 g 1 + ni=2 F β A2i ĝ i ,
i
(9.1696)
exp(2ϕ) = A2ϕ F βϕ , (9.1697)
CBPF-MO-003/02 240
and the parameters β i ( i ¿1 ), βϕ satisfy the relations (9.1690). The special case of the solution
(9.1690), (9.1696)-(9.1698) with βϕ = 0 (constant scalar field) was obtained earlier in [61].
where t > 0, −∞ < xi < ∞ and αi are constants, i = 1, . . . , n. From Appendix 1 we get
where n n
X X
F (α) = [2αi2 (αi − 1)2 − αi4 ] + [ αi2 ]2 . (9.1701)
i=1 i=1
The metric (9.1699) with the restrictions (9.1702) imposed satisfies the vacuum Einstein equations
(or, equivalently, RM N [g] = 0). It is a trivial generalization of the well-known Kasner solution. In
this case n
X
F (α) = Φ(α) = Φn (α) ≡ [αi4 − 4αi3 ] + 3. (9.1703)
i=1
CBPF-MO-003/02 241
where n n
X X
n
E = En ≡ {α = (α1 , . . . αn ) ∈ R | αi = αi2 = 1}. (9.1705)
i=1 i=1
n−2
Notice, that E is (n − 2)-dimensional ellipsoid for n > 2 (E ' S ).
For n = 1, M = E = {(1)} and we are lead to well-known Milne solution
We recall that by the coordinate transformation y 0 = t cosh x1 , y 1 = t sinh x1 the metric (9.1706) is
reduced to the Minkowsky metric η = −dy 0 ⊗ dy 0 + dy 1 ⊗ dy 1 in the upper light cone y 0 > |y 1 |.
For α = (. . . , 0, 1i , 0, . . .) ∈ M we get a trivial extension of the Milne metric:
n
X
2 i i
gm = −dt ⊗ dt + t dx ⊗ dx + dxj ⊗ dxj , (9.1707)
j6=i
i = 1, . . . , n; n > 1.
Proposition 9.2.1. Let α = (α1 , . . . αn ) ∈ E. Then Φ(α) ≥ 0 and Φ(α) = 0 if and only if
α ∈ M.
Proof. For n = 1, 2 the proposition is trivial. So, we consider the case n > 2. Let
Φ| = Φ|E : E −→ R (9.1708)
where µ, λ ∈ R. The point α belongs to Ext if and only if there exist λ, µ ∈ R such that (α, µ, λ) is
a point of extremum of the function (9.1709), i.e. the relations (9.1702) and
∂ Φ̂
= 4αi3 − 12αi2 + 2µαi + λ = 0, (9.1710)
∂αi
i = 1, . . . , n, are satisfied. From (9.1702) and (9.1710) we obtain
n
X
4 αi3 − 12 + 2µ + λn = 0, (9.1711)
i=1
Xn n
X
4 αi4 − 12 αi3 + 2µ + λ = 0, (9.1712)
i=1 i=1
CBPF-MO-003/02 242
and hence
4Φ(α) = λ(n − 1). (9.1713)
Let us consider the cubic equation
We prove that for given µ and λ the cubic equation (9.1714) should have three different real roots.
Indeed, there should be at least two different real roots since otherwise α1 = . . . = αn but this
is impossible due to the Kasner constraints (9.1702). The third root should be also real and we
are lead to the following three possibilities for the roots: i) y1 = y2 < y3 ; ii) y1 < y2 = y3 ; iii)
y1 < y2 < y3 . It follows from (9.1714) that
y1 + y2 + y3 = 3 (9.1715)
and hence y3 > 1. But due to (9.1702) αi ≤ 1 and as a consequence the possibilities i) and ii) can
not be fulfilled (otherwise αi = y1 for all i). Thus y1 < y2 < y3 and due to (9.1710) and y3 > 1 we
obtain
{x|x = αi , i = 1, . . . , n} = {y1 , y2 }. (9.1716)
Solving (9.1702) for α satisfying (9.1716) we get
√
m1 − ∆
y1 = y1 (m1 , m2 ) = , (9.1717)
m1 n
√
m2 + ∆
y2 = y2 (m1 , m2 ) = , (9.1718)
m2 n
∆ = m1 m2 (n − 1). (9.1719)
Ext = E1 t . . . t En , (9.1722)
where
Φ(α) = 0. (9.1725)
CBPF-MO-003/02 243
Using the inclusion Min ⊂ Ext and the relations (9.1722), (9.1724) and (9.1725) we obtain
The proposition 9.2.1 follows from the relations (9.1725) and (9.1726).
Remark. From (9.1700), (9.1703) and Proposition 9.2.1 we get for the Kasner metric (9.1699)
with α ∈ E \ M
I[g](t, ~x(t)) → +∞, as t → +0. (9.1727)
for arbitrary curve ~x(t).
The metric (9.1682) with the restrictions (9.1728) imposed satisfies the vacuum Einstein equations
[61].
For the metric (9.1682), (9.1728) we get from Appendix 1
n
X
I[g] = t−4αi c−2 i −4
i I[g ] + 2Φ∗ (α)t , (9.1729)
i=1
where n
X
Φ∗ (α) ≡ di [αi4 − 4αi3 ] + 3. (9.1730)
i=1
where n n
X X
n
E∗ ≡ {α = (α1 , . . . αn ) ∈ R | di αi = di αi2 = 1}. (9.1732)
i=1 i=1
β1 = . . . = βd1 = α1 ,
...
βN −dn +1 = . . . = βN = αn . (9.1734)
CBPF-MO-003/02 244
It is evident that β ∈ EN (see (9.1705)). The Proposition 9.2.2 follows from the Proposition 9.2.1,
relation (9.1733) and the equivalence
β(α) ∈ M ⇐⇒ α ∈ M∗ . (9.1735)
I[g i ] ≥ 0, (9.1736)
as τ → +0, where ci 6= 0, αi are constants, i = 1, . . . , n. (We remind that the notation ϕ(τ ) = o(1)
as τ → +0 means that ϕ(τ ) → 0 as τ → +0.)
Remark. The relations (9.1744) and (9.1745) should not obviously follow from (9.1743). A
simple counterexample is
1
Ai (τ ) = 1 + τ sin 2 . (9.1746)
τ
But if
Ai (τ ) = ci τ 2αi [1 + ϕi (τ )], (9.1747)
where ci 6= 0 and
ϕi (τ ) = o(1), τ ϕ̇i (τ ) = o(1), τ 2 ϕ̈i (τ ) = o(1), (9.1748)
as τ → +0, i = 1, . . . , n, then the relations (9.1743)-(9.1745) are satisfied.
Theorem. Let g be the metric (9.1739) defined on the manifold (9.1740), where the metrics g i ,
i = 1, . . . , n, satisfy the relations (9.1741) and (9.1742). Let the scale factors Ai (τ ), i = 1, . . . , n,
satisfy the relations (9.1743)-(9.1745), where the set of parameters α = (αi ) satisfies the Kasner-like
relations (9.1728) (di = dimMi ) and is non-exceptional, i.e. α ∈ / M∗ (M∗ is defined in (9.1731)).
Then
I[g](τ, x) → +∞, as τ → +0, (9.1749)
uniformly on x ∈ M1 × . . . × Mn .
Proof. From Appendix 1 we get
where
n
X
I1 [g] = A−2 i
i I[g ], (9.1751)
i=1
Xn
I2 [g] = A−3 2 i
i Ȧi wR[g ], (9.1752)
i=1
n
X 1 1
I3 [g] = {− di A−4 4 −1 −2 2 2
i Ȧi + di (2Ai Äi − Ai Ȧi ) }
i=1
8 4
n
1X
+ [ di (A−1 2 2
i Ȧi ) ] . (9.1753)
8 i=1
Ȧ2i
= c−1
i τ
−2−2αi
[4αi2 + o(1)], (9.1756)
A3i
CBPF-MO-003/02 246
αi < 1, (9.1757)
i = 1, . . . , n. Let
δ = min(1 − αi ) > 0. (9.1758)
i
Then it follows from (9.1756) that there exists τ1 > 0 such that
Ȧ2i
0≤ < τ −4+δ , (9.1759)
A3i
Ȧ2i
3
wR[g i ](xi ) ≥ −|Di |τ −4+δ (9.1760)
Ai
A−2 −2 −4αi
i = ci τ [1 + o(1)], (9.1762)
A−2
i < τ
−4+δ
(9.1763)
for all τ < τ2 , i = 1, . . . , n. Using (9.1741) and (9.1763) we obtain (analogously to (9.1761))
The metric and scalar field from (10.1766), (10.1767) satisfy the field equations (9.1678), (9.1679).
This solution is a scalar-vacuum multispace generalization of the Tangherlini solution [96]. It may
be obtained as a special case of the solution (9.1696), (9.1697) by certain analitical continuation in
R-variable and R0 -parameter. In the parametrization of the harmonic-type variable this solution
was presented earlier in [100, 102].
For aϕ = 0 this solution was obtained in [98, 102]. Some special solutions were (with aϕ = 0 )
were considered earlier in the following publications: [93]-[94] (d = 2; n̄ = 1; d1 = 1), [315, 24] (d =
2; n̄ = 2, 3; d1 = . . . = dn̄ = 1), [301] (d = 2; n̄ = 1), [302] ( n̄ = 1; a = [(1 − (d + d1 )−1 )/(1 − d−1 )]1/2 ;
a1 = −a/(d + d1 − 1)); [303] (n̄ = 2, d2 = 1), [97] (n = 1 and d = 2), [304] (d = 2; n̄ is arbitrary).
We also note that the cosmological analogue of the solution [98] was presented in [61], where
the tree-generalizations of the solution were considered. (Such tree-generalizations may be also
constructed for spherically-symmetric case).
Proposition 10.1.1. The (2 + d)-dimensional section of the metric (10.1766) has a horizon at
R = R0 if and only if
a = 1, a1 = . . . an = aϕ = 0. (10.1772)
In the special case aϕ = 0 this proposition was proved in [98] (for d = 2 see also [304]).
Proposition 10.1.2. Let g be the metric (10.1775)-(10.1777) with the parameters (10.1783),
(10.1784) satisfying the relations (10.1785), and obeying the restrictions: αϕ = 0, α = (α−1 , . . . αn ) ∈
/
i
M1 . Let Ricci-flat internal spaces (Mi , g ), i = 1, . . . , n, satisfy the self-boundness conditions
(9.1741). Then
I[g](τ, y) → +∞ as τ → +0, (10.1789)
uniformly on y ∈ R × S d × M1 × . . . × Mn .
Proof. We denote (M−1 , g (−1) ) = (R, −dt ⊗ dt) and (M0 , g (0) ) = (S d , dΩ2d ). Due to assumption
of the proposition, flatness of M−1 -space and the relations
the conditions of the Theorem ((9.1741), (9.1742)) are satisfied for all spaces (Mν , g ν ), ν = −1, . . . , n.
All scale factors Aν (τ ), ν = −1, . . . , n, and their first and second derivatives satisfy the Kasner-
like asymptotical conditions of the theorem (see (9.1743)-(9.1745)). This may be proved using
asymptotical relations (10.1780), (10.1781) and (10.1773). Thus, the Proposition 10.1.2 follows
from the Theorem ¿from the previous section.
Using equivalence (10.1788) and we may reformulate the Proposition 10.1.2 for the metric in the
representation (10.1766).
Proposition 10.1.3. Let g be the metric (10.1766) with the parameters satisfying (10.1770),
(10.1771) and aϕ = 0, (a, a1 , . . . an ) ∈ / T (see (10.1787)). Let Ricci-flat internal spaces (Mi , g i ),
i = 1, . . . , n, satisfy the self-boundness conditions (9.1741). Then
I[g](R, y) → +∞ as R → R0 , (10.1791)
uniformly on y ∈ R × S d × M1 × . . . × Mn .
Remark . Due to Appendix 1, (10.1790) and the flatness of t-space I[g] does not depend on
yj ∈ Mj , j = −1, 0.
Remark . From (9.1678) we obtain
we obtain for aϕ 6= 0
R[g](R, y) → +∞, as R → R0 . (10.1794)
Let us consider the special case of the solution (10.1766)-(10.1768) with n − 1 one-dimensional
internal spaces (extra times). This solution defined on the manifold
reads
n
X
g= − f ai dti ⊗ dti + f b−1 dR ⊗ dR + f b R2 dΩ2d , (10.1796)
i=1
exp(2ϕ) = Bϕ f aϕ , (10.1797)
f = f (R) = 1 − (R0 /R)d−1 , where Bϕ , R0 > 0 are constants and the parameters b, a1 , . . . , an satisfy
the relations
n
X
b = (1 − ai )/(d − 1), (10.1798)
i=1
Xn n
X
2 2
( ai ) + (d − 1)(aϕ + a2i ) = d. (10.1799)
i=1 i=1
Let us consider the case aϕ = 0 [102]. The ”Tangherlini set” in this case (see (10.1787))
consists of n points.
We call the points Ti as Tangherlini points and the set (10.1800) as Tangherlini set. The metric
(10.1796) for a = Tk has a rather simple form
(k)
X
g = gT − dti ⊗ dti , (10.1801)
i6=k
(k)
where gT is the Tangherlini solution with the time variable t = tk , k = 1, ..., n. The metric
(10.1801) is a trivial (cylindrical) extension of the Tangherlini solution with the time tk . It describes
an extended membrane-like (string-like for n = 2) object. Any section of this object by hypersurface
ti = ti0 = const, i 6= k, is the (2 + d)-dimensional black hole [96, 284], ”living” in the time tk .
As was shown in [102] (see Proposition 10.1.4 below) the multitemporal horizon takes place only
for (a1 , . . . , an ) ∈ T . For (a1 , . . . , an ) ∈
/ T we get from Proposition 9.2.5
i.e. the solution (10.1796) describes a multitemporal naked singularity. (The relation (10.1802) was
stated previously in [102].)
Remark. This singular solution is unstable under monopole perturbations: this follows from
the result of Bronnikov et al [189, 100, 285].
When aϕ 6= 0, we get from (10.1794)
ẍM + ΓM N P
N P [g]ẋ ẋ = 0. (10.1804)
κ = dθ1 ⊗ dθ1 + sin2 θ1 dθ2 ⊗ dθ2 + . . . + sin2 θ1 . . . sin2 θd−1 dθd ⊗ dθd (10.1806)
f ai ṫi = εi , (10.1807)
f b R2 ϕ̇ = j, (10.1808)
n
X
b−1 2 2 −b −2
f (Ṙ) + j f R − (εi )2 f −ai = 2ER (10.1809)
i=1
i = 1, ..., n. We put here θ1 = . . . = θd−1 = π2 . (This may be done for any trajectory by a suitable
choice of coordinate system.) The radial equation
n
j 2 −b −2 0 1 1 X i 2 −ai 0
(f b−1 Ṙ). + (f R ) − (Ṙ)2 (f b−1 )0 − (ε ) (f ) = 0 (10.1810)
2 2 2 i=1
We note, that the case 2ER = 2LR > 0 in (10.1809) correspond to a tachion.
Multitemporal horizon. Here we consider the null geodesics. Putting ER = 0 in (10.1809)
we get for a light ”moving” to the center
v
u n
uX
Ṙ = −t (εi )2 f 1−b−ai − j 2 f 1−2b R−2 (10.1812)
i=1
and consequently
Z R
i εi [f (x)]−ai
t − ti0 =− dx pPn , (10.1813)
i 2 1−b−ai − j 2 [f (x)]1−2b x−2
R0 i=1 (ε ) [f (x)]
CBPF-MO-003/02 252
i = 1, . . . , n.
Definition. We say that the ε-horizon takes place for the metric under consideration at R = R0
if and only if
n
X
||t − t0 || ≡ |ti − ti0 | → +∞, (10.1814)
i=1
Indeed, the integrand in the i-th integral in (10.1813) behaves like εi (R0 − x)si as x → R0 . The set
of inequilities (10.1815) may be rewritten as following
for all i ∈ Kε , where maxε in defined analogously to minε . It can be easily verified that the set of
inequalities (10.1817) is equivalent to the following inequality
d(a1 − 1) + a2 + . . . + an = 0. (10.1820)
or, equivalently,
a1 < 1 + b. (10.1822)
In analogous manner it may be proved that
ai < 1 + b. (10.1823)
for all a ∈ E \ Ti , i = 1, ..., n. The inequalities (10.1823) imply (10.1819). The proposition is proved.
Now we consider the case a ∈ T . Without loss of generality we put a = T1 = (1, 0, . . . , 0). It is
not difficult to verify that in this case the ε-horizon takes place only if ε1 6= 0.
CBPF-MO-003/02 253
coincide with the geodesic equations (10.1804). In this case (E i ) = (mεi ) is the energy vector
and J = mj is the angular momentum (see (10.1807) and (10.1808)). For fixed values of εi the
(d+1)-dimensional part of the equations of motion is generated by the Lagrangian
X n
m b
L∗ = [f ḡT,αβ (x)ẋα ẋβ + (εi )2 f −ai ], (10.1826)
2 i=1
tr(M MI )
V = −G (10.1829)
Rd−1
where MI = (mεi εj ) is the inertial mass matrix of the particle. Thus, we obtained a generalization
of the Newton law for multitemporal case.
The solution (10.1796) may be also rewritten in the matrix form
and
b = (1 − trA)/(d − 1). (10.1832)
CBPF-MO-003/02 254
Here xA ≡ exp(A ln x) for x > 0. The metric (10.1830) can be reduced to the metric (10.1796) by
the diagonalization of the A-matrix: A = S T (ai δij )S, S T S = 1n and the reparametrization of the
time variables: t̄i = Sij tj . In this case the gravitational mass matrix is
(Mij ) = (Aij B/2G). (10.1833)
We may also define the gravitational mass tensor as
M = Mij dt̄i ⊗ dt̄j . (10.1834)
We call the extended object, corresponding to the solution (10.1830)-(10.1832) as ”multitemporal
hedgehog”. At large distances Rd−1 À B this object is described by the matrix analogue of the
Newton’s potential
1
Φij = − BR1−d Aij = −GR1−d Mij . (10.1835)
2
Clearly that this potential for the diagonal case (10.1796) A = ai δij is a superposition of the
potentials, corresponding to ”pure” black hole states (10.1801).
Remark. It is interesting to note that the formula
A = Qi R1−d dti (10.1836)
describe the multitemporal O(d+1)-analogue of the well-known electrostatic solution of the Maxwell
equations. In this case the charge Q = (Qi ) is a vector (or we may also define the charge as the
1-form Qi dti ).
Remark. Let us consider the solution (10.1796) for n = 2 with a1 > 0 and a2 < 0. In this case
under a suitable choice of the εi -parameters a point R > R0 , may be a libration point, i. e. the
point of equilibrium. In this case
a1 (ε1 )2 + a2 (ε2 )2 [f (R)]a2 −a1 = 0 (10.1837)
and ε2 6= 0. An analogous situation takes place for arbitrary n, when there exist positive and
negative ai -th parameters.
where g = gM N dxM ⊗ dxN is the metric , F = 21 FM N dxM ∧ dxN = dA is the strength of the
electromagnetic field and ϕ is the scalar field (dilatonic field). Here λ is the dilatonic coupling
constant and κ denotes gravitational constant. The action (11.1840) describes for certain values of
the coupling constant λ and space-time dimension D a lot of interesting physical models including
standard Kaluza-Klein theory, certain sectors of supergravity theories etc. For
1
λ2 = , D = 10, (11.1841)
8
the action (11.1840) describes a part of the bosonic sector for the N = 1 ten-dimensional Einstein-
Yang-Mills supergravity that occurs in the low energy limit of superstring theory [31].
In [284] Myers and Perry obtained the multidimensional O(d + 1)-symmetric analogue of the
well-known Reissner-Nordström charged black hole solution. In [285, 102] the generalization of the
Myers-Perry solution to the case of n Ricci-flat internal spaces was obtained. Special cases of the
solution [285, 102] were considered earlier in the following publications: [99, 59, 286] (d = 2, λ = 0),
[287] (d = 2, n = 1), [288] (d ≥ 2, λ = 0). (In [98] the special case of solution [285, 102] with zero
electric and scalar charges was obtained.)
Here we consider as special case of the solution [285, 102] a charged dilatonic black hole with n
internal Ricci-flat spaces. Objects of such sort are very popular in the literature (see, e.g. [285, 287,
49, 289, 202, 290, 291]. Here we present bounds on the mass of this black hole and the Hawking
temperature. For D = 4, d = 2 the mass bound was obtained by Gibbons and Wells in [291]. We
also consider a charged black hole solution with infinite number of internal spaces n = ∞. We note
that the Einstein equations in the infinite-dimensional space were considered earlier by Kalitzin
[292]. Infinite-dimensional spherically-symmetric and cosmological solutions were also presented in
[102] and [61], respectively.
where
Here we consider the spherically O(d + 1)-symmetric solutions to the field equations (11.1842)-
(11.1844) obtained in [285, 102]. These solutions are special case of those considered in Chapter 4.
The solution [102] is defined on the manifold
M = M (2+d) × M1 × . . . × Mn , (11.1846)
C2 d C1 (D − 2)
= 2
+ B 2 (1 + λ2 )
d−1 D − 3 + λ (D − 2)
Xn Xn
1
+ (λB + Ai di )2 + A2i di . (11.1855)
d−1 i=1 i=1
κ2 Q2 A(λ)
B+ B− = . (11.1866)
(d − 1)2 (D − 2)
In this case the (2 + d)-dimensional section of the metric (11.1861) has a horizon at rd−1 = B+ .
For rd−1 = B− < B+ the horizon is absent (λ 6= 0) since
αt − αr < 0. (11.1868)
where
D − 3 − λ2 (D − 2)
β(λ) = 1 + 2αt = (11.1870)
D − 3 + λ2 (D − 2)
CBPF-MO-003/02 258
κ|Q|
Mc = √ . (11.1878)
G(d − 1) 1 + λ2
corresponds to the tree-level string effective action in D dimensions. For the string case (11.1881)
α = αs = (d − 2)/(d − 1) ≥ 0. From (11.1880) we get the inequality
(d − 1) ³ B− ´
TH > T M P = 1− (11.1882)
4πr+ B+
where TM P is the Hawking temperature for the Myers-Perry charged black hole [166] (λ = 0,
D = d + 2). For D = 4 (d = 2) relation (11.1879) agrees with the corresponding relation from [290].
For the Hawking temperature (11.1879) we get in the case under consideration
(d − 1) ³ B− ´ d−2
d−1
TH = TH,s = 1− . (11.1883)
4πr+ B+
Thus, in the string case (11.1881) the Hawking temperature (11.1883) does not depend upon the
total dimension D, or equivalently upon the internal space dimension dint . Moreover, it does not
depend upon the Ricci-flat internal space (Mint , gint ). For example, one may consider for D = 10
different internal Calabi-Yau spaces [31] with the same result for TH . For d = 2 (11.1883) coincides
with the corresponding formula for the Schwarzschild black hole.
Our solution to the field equations is defined on the manifold (11.1888) and has the following
form
hX
1+d i
(3−D)2/A(λ) 2/A(λ) a a
g = −U dt ⊗ dt + U dx ⊗ dx + ĝint , (11.1889)
a=1
exp(2λϕ) = U −2αt , (11.1890)
νdt
A = AM dxM = , (11.1891)
κU
where αt is defined in (11.1865), F = dA,
ν 2 = (D − 2)/A(λ), (11.1892)
U = U (t, ~x) = ht + Φ(~x), (11.1893)
4Φ = δ ab ∂a ∂b Φ = 0, (11.1894)
for Λ = 0.
Special cases of the above solutions with Λ > 0 were also considered by Kastor and Traschen
[295] (D = 4, d = 2, λ = 0) and Horne and Horowitz [296] (D = 4, λ2 = 1/2). The solution [295]
generalizes the well-known Majumdar-Papapetrou solution [166]. For Λ 6= 0 and
m
X Bi
Φ=1+ (11.1900)
i=1
|~x − ~xi |d−1
It should be also noted that the global properties of static spherically symmetric solutions to the
Einstein-Maxwell-dilaton system in the presence of an arbitrary exponential dilaton potential were
derived in [299].
Extremal charged black hole for Λ = 0.
Now we consider the solution (11.1861)-(11.1866) in the extremal case B+ = B− = B, where
p
κ|Q| A(λ)
B= p , (11.1901)
(d − 1) (D − 2)
and denoting
B
U =1+ = f±−1 (11.1903)
Rd−1
we get a special case of the solution with Λ = h = 0 and m = 1 in (11.1900) (R = |~x − ~x1 |, B = B1 ).
(See also Chapter 5.)
In the extremal case a horizon for rd−1 = B, takes place if (and only if)
α = α(λ, d, D) ≥ 0 (11.1904)
(see (11.1880)). This is equivalent to the following restriction on the dilatonic coupling parameter
1
λ2 ≤ d − 2 + ≡ λ2c . (11.1905)
(D − 2)
For these values of λ the inequality (11.1872) should be replaced by the relation: M ≥ Mc .
The relation (11.1905) is satisfied for the string case (11.1881). For d = 2 we have λ2c = λ2s .
For λ2 = λ2c we get α = 0 and
d−1
TH = Tc ≡ . (11.1906)
4πr+
Due to relation (11.1879) the Hawking temperature has the following limit as B− → B+ , i.e. in
the extreme black hole limit
and the relations (11.1865), (11.1892) and (11.1895) take the following form
Thus, the spatial section of the metric (11.1909) is flat. For string (11.1881), “Kaluza-Klein”
(11.1885) and critical (11.1905) values of the coupling constants we have, respectively,
Stringy case. For λ2 = λ2s we obtain instead of (11.1909)-(11.1911) the following relations
hX
1+d i
(3−D)2/(D−2) 2/(D−2)
g = −U dt ⊗ dt + U dxa ⊗ dxa + ĝint , (11.1913)
a=1
12 Appendix
12.1 Appendix 1
12.1.1 Ricci-tensor components
The nonzero Ricci tensor components for the metric (7.1415) are the following [118]
h n
X
Rµν [g] = Rµν [g 0 ] + gµν
0
−∆0 γ + (2 − d0 )(∂γ)2 − ∂γ dj ∂φj ] (12.1914)
j=1
n
X
+ (2 − d0 )(γ;µν − γ,µ γ,ν ) − di (φi;µν − φi,µ γ,ν − φi,ν γ,µ + φi,µ φi,ν ), (12.1915)
i=1
i
n n
X o
i
Rmi ni [g] = Rmi ni [g ] − e2φ −2γ gm
i
i ni
i i
∆0 φ + (∂φ )[(d0 − 2)∂γ + dj ∂φj ] , (12.1916)
j=1
0 µν
Here ∂β ∂γ ≡ g β,µ γ,ν and ∆0 = ∆[g ] is the Laplace-Beltrami operator corresponding to g 0 and
0
all covariant derivatives correspond to g 0 . The scalar curvature for (7.1415) is [118]
Xn n n
X
−2φi i −2γ 0
R[g] = e R[g ] + e R[g ] − di (∂φi )2 (12.1917)
i=1 i=1
o
− (d0 −2)(∂γ)2 − (∂f )2 − 2∆0 (f + γ) , (12.1918)
where f is defined in (2.73). Relations for the Ricci tensor may be obtained using the relations for
the Riemann tensor from the next subsection and the relations for the conformal transformations
from the last subsection.
i = 1, . . . , n.
where
X d(Is )
α = α(b) ≡ (d0 − 2) (−εs )νs2= (d0 − 2)η(b) + 1, (12.1936)
D−2
s∈S(b)
X · ¸
2 d(Is )
αi = αi (b) ≡ (d0 − 2) (−εs )νs δiIs − , (12.1937)
D−2
s∈S(b)
i = 1, . . . , n.
It follows from definitions (??)-(12.1935) that C ≥ 0 and
C = 0 ⇔ (α = 0, 2, αi = 0, i = 1, . . . , n). (12.1938)
It is clearly, that
(F1 · F2 )M N = (F2 · F1 )N M , F1 F2 = F2 F1 . (12.1947)
For the volume forms (7.1316) we get
where indices mi , ni correspond to the manifold Mi , i = 1, . . . , n. The symbols ε(I) and δiI are
defined in (7.1317) and (2.41) respectively.
For the form F (a,e,I) from (??) and metric g from (7.1415) we obtain from (12.1948)-(12.1949)
1 A(I)
(F (a,e,I) · F (a,e,I) )µν = ∂µ Φ(a,e,I) ∂ν Φ(a,e,I) exp(2γ); (12.1950)
na ! na
1 A(I)
(F (a,e,I) · F (a,e,I) )mi ni = δiI gm
i
i ni
(∂Φ(a,e,I) )2 exp(2φi ), (12.1951)
na ! na
where indices mi , ni correspond to the manifold Mi , i = 1, . . . , n, and
³ X ´
i
A(I) = ε(I) exp −2γ − 2 di φ , (12.1952)
i∈I
I ∈ Ωa,e . All other components of (F (a,e,I) · F (a,e,I) )M N are zero. For the scalar invariant we have
1 1 (a,e,I) (a,e,I)
(F (a,e,I) )2 ≡ F F = A(I)(∂Φ(a,e,I) )2 , (12.1953)
na ! na !
I ∈ Ωa,e . Here, as above, we use the notations: ∂Φ1 ∂Φ2 = g 0µν ∂µ Φ1 ∂ν Φ2 and (∂Φ1 )2 = ∂Φ1 ∂Φ1 for
functions Φ1 = Φ1 (x) and Φ2 = Φ2 (x) on M0 .
Analogous relations for magnetic case may be obtained using the formulas
1 ε[g]
(∗F1 )(∗F2 ) = F1 F2 , (12.1954)
k∗ ! k!
1 ε[g]
[(∗F1 ) · (∗F2 )]M N = {gM N (F1 F2 ) − k(F2 · F1 )M N }, (12.1955)
(k∗ − 1)! k!
where k = rank Fi and k∗ = rank(∗Fi ) = D − k, i = 1, 2.
Let I, J ∈ Ω, I 6= J and d(I) = d(J). Then
The last term in (12.1961) gives rise to off-block-diagonal components of stress-energy tensor from
(2.54).
For a ∈ ∆ (d0 6= 2) we obtain
F (a,e,I) F (a,m,J) = F (a,m,J) F (a,e,I) = 0, (12.1962)
I ∈ Ωa,e , J ∈ Ωa,m , and hence
F a,e F a,m = F a,m F a,e = 0, (12.1963)
(F a )2 = (F a,e )2 + (F a,m )2 , (12.1964)
where F a = F a,e + F a,m . We also get
(F a · F a )M N = (F a,e · F a,e )M N + (F a,m · F a,m )M N
(12.1965)
a,e a,m a,m a,e
+(F ·F )M N + (F ·F )M N (12.1966)
for a ∈ ∆. The last two terms in (12.1965) give rise to off-block-diagonal components of stress-energy
tensor from (2.61) and (2.62).
In these cases we have the following formulas for inverse Cartan matrices
( (
0
ss0 min(s, s ) for s =
6 r, ss0 min(s, s0 ) for s0 6= r,
A = 1 0
A = 1
(12.1972)
2
s for s = r, 2
s for s0 = r
and ½
s(2r + 1 − s) for s 6= r,
ns = r ns = s(2r − s), (12.1973)
2
(r + 1) for s = r;
for Br and Cr series respectively, s = 1, . . . , r.
Dr series. We have the following Dynkin diagram for this case (Fig. A.3):
r
u
¡
u u u ... u
¡
@
1 2 3 r − 2 @u
r−1
Then ½
s(2r − 1 − s) for s ∈
/ {r, r − 1},
ns = r (12.1975)
2
(r − 1) for s ∈ {r, r − 1},
s = 1, . . . , r.
Let us consider the exceptional Lie algebras. Dynkin diagrams of these algebras are pictured on
Fig. 4.A.
u6 u7
u u u u u u u u u u u
1 2 3 4 5 1 2 3 4 5 6
CBPF-MO-003/02 269
u8
u u u u u u u u u> u u u> u
1 2 3 4 5 6 7 1 2 3 4 1 2
for all s ∈ S.
Then, the Euler-Lagrange equations for the Lagrangian (12.1977)
X
ẍ + 2As us exp(2 < us , x >) = 0, (12.1979)
s∈S
with
2 < us , us0 >
Ass0 = , (12.1983)
< u s0 , u s0 >
CBPF-MO-003/02 270
s, s0 ∈ S. Let the matrix (Ass0 ) be a non-degenerate one. In this case vectors us , s ∈ S, are linearly
independent. Then eqs. (12.1982) are field equations corresponding to the Lagrangian
1 X −1 X X
Ks Ass0 q˙s q˙s0 −
0
LT L = As exp( Ass0 q s ). (12.1984)
4 s,s0 ∈S s∈S s0 ∈S
where
1 X −1 X X
Ks Ass0 q˙s q˙s0 +
0
ET L = As exp( Ass0 q s ), (12.1986)
4 s,s0 ∈S s∈S s0 ∈S
s ∈ S.
S = S1 t . . . t Sk , (12.1988)
all Si 6= ∅, and
< us , us0 >= 0, (12.1989)
for all s ∈ Si , s0 ∈ Sj , i 6= j; i, j = 1, . . . , k.
0
Let hs = Ks−1 , (Ass ) = (Ass0 )−1 , X 0
bs = 2 Ass , (12.1990)
s0 ∈S
s, s0 ∈ Si , i = 1, . . . , k. More explicitly
s, s0 ∈ Si , i = 1, . . . , k, and
1 √
s(t, ξ, C) ≡ √ sh(t C), ξ = +1, C > 0; (12.1998)
C
1 √
√ sin(t −C), ξ = +1, C < 0; (12.1999)
−C
t, ξ = +1, C = 0; (12.2000)
1 √
√ ch(t C), ξ = −1, C > 0. (12.2001)
C
For ”Toda” part of energy we get
1X
ET L = Cs bs hs . (12.2002)
2 s∈S
∂2
2Ĥ0 = − 2 + 2A e2qz , (12.2003)
µ ∂z¶
∂ ∂
2Ĥ1 = − eqz e−qz + 2A e2qz . (12.2004)
∂z ∂z
Equation
Hk Ψk = EΨk (12.2005)
has the following linearly independent solutions for q 6= 0
µ ¶
kqz/2
√ eqz
Ψk (z) = e Bωk (E) 2A , (12.2006)
q
s
k 2E
ωk (E) = − 2, (12.2007)
4 q
∂A vB + ∂B vA = 0, (12.2012)
∂A vs + ∂s vA − 2UAs vs = 0, (no summation) (12.2013)
∂s vs0 + ∂s0 vs = 0, s 6= s0 (12.2014)
s
∂s vs + εs e2U UAs v A = 0. (12.2015)
where CAB (Φ) = −CBA (Φ), A, B = 1, . . . , N . The Proposition 4 may be readily proved using the
relations (12.2013)-(12.2015), identities ∂M ∂N (vs exp (−2UAs xA )) = ∂N ∂M (vs exp (−2UAs xA )) and the
following Lemmas.
Lemma 1. Let (UA ) ∈ RN , (UA ) 6= 0 and
A
D = eUA x (B + CA xA ) (12.2017)
ηA = ±1; A, B = 0, . . . , D − 1.
We choose the following frame vectors
1 n
(eA M ) = diag(eγ e(0)aµ , eφ e(1)a1m1 , . . . , eφ e(n)am
n
n
), (12.2020)
where
0 (0) (i)
gµν = ηab e(0)aµ e(0)b ν , i
gm i ni
= ηai bi e(i)aimi e(i)bini , (12.2021)
i = 1, . . . , n, and
(0) (1) (n)
(ηAB ) = diag(ηab , ηa1 b1 , . . . , ηan bn ). (12.2022)
For (eMA ) = (eA M )−1 we get
1 n
(eMA ) = diag(e−γ e(0)µa , e−φ e(1)ma11 , . . . , e−φ e(n)mann ), (12.2023)
(0)µ (0)a (i)m (i)a
where (e a ) = (e µ )−1 , (e ai i ) = (e ini )−1 , i = 1, . . . , n.
Indices. For indices we also use an alternative numbering: A = (a, a1 , . . . , an ), B = (b, b1 , . . . , bn ),
where a, b = 10 , . . . , (d0 )0 ; a1 , b1 = 11 , . . . , (d1 )1 ; ...; an , bn = 1n , . . . , (dn )n ; and M = (µ, m1 , . . . , mn ),
N = (ν, n1 , . . . , nn ), where µ, ν = 10 , . . . , (d0 )0 ; m1 , n1 = 11 , . . . , (d1 )1 ; ...; mn , nn = 1n , . . . , (dn )n .
12.7.2 Gamma-matrices
In what follows Γ̂A are ”frame” Γ-matrices satisfying
where the covariant derivative 5M [g] corresponds to the metric g = g(e, η) from (12.2019). The
spinorial covariant derivative reads
1
DM = ∂M + ωABM Γ̂A Γ̂B , (12.2027)
4
0
where ωABM = ηAA0 ω A BM .
CBPF-MO-003/02 274
The non-zero components of the spin connection (12.2026) in the frame (12.2020) read
(0) (0)ν
ω a bµ = ω a bµ (e(0) , η (0) ) − e(0)νa γ,ν e bµ +e b γ,ν e
(0)a
µ, (12.2028)
φi −γ
ω a ai mj = −δij e (e(0)aν 5ν [g (0) ]φi )e(i)ai mi , (12.2029)
φi −γ
ω aiamj = δij e (e(0)ν a ∂ν φi )e(i)aimi , (12.2030)
ω aibj mk = δij δjk ω aibi mi (e(i) , η (i) ), (12.2031)
i, j, k = 1, . . . , n, where ω a bµ (e(0) , η (0) ) and ω aibi mi (e(i) , η (i) ), are components of the spin connections
corresponding to the metrics from (12.2021).
Let
AM ≡ ωABM Γ̂A Γ̂B . (12.2032)
For AM = AM (e, η, Γ̂) in the frame (12.2020) we get
(0)
Aµ = ωabµ Γ̂a Γ̂b + (Γµ Γν − Γν Γµ )γ,ν , (12.2033)
(i)
Ami = ωai bi mi Γ̂ai Γ̂bi + 2Γmi Γν φi,ν , (12.2034)
(0) (i)
where ωabµ = ωabµ (e(0) , η (0) ) and ωai bi mi = ωai bi mi (e(i) , η (i) ), i = 1, . . . , n.
Q (i)m
where l = 0, . . . , n, with m0 = µ, s(I) = sign( i∈I det(e ai i )) and Γ̂(I) = Γ̂1̄ Γ̂2̄ Γ̂3̄ with (1̄, 2̄, 3̄) =
(1i1 , . . . , (di1 )i1 , . . . , 1ik , . . . , (dik )ik ).
M 5-brane. Let
¯
F = (∗0 dΦ) ∧ τ (I) (12.2040)
where ∗0 is the Hodge operator on (M0 , g 0 ) and I¯ = {1, . . . , n} \ I = {j1 , . . . , jl }, j1 < . . . < jl . It
follows from (12.2040) that d0 + d(I) ¯ = 5 and d(I) = 6. We get
1 X
Bml = ¯ exp[−(d0 − 2)γ −
√ s({0})s(I) di φi ] × (12.2041)
12 2 i∈I¯
¯
× [2Γml Γν Φ,ν − 3δ0l (Γml Γν − Γν Γml )Φ,ν + 6δIl¯Γν Γml )Φ,ν ]Γ̂({0})Γ̂(I), (12.2042)
(0)ν ¯ = Γ̂1̄ . . . Γ̂k̄ with
where l = 0, . . . , n; s({0}) = sign(det(e a )), Γ̂({0}) = Γ̂10 . . . Γ̂(d0 )0 and Γ̂(I)
(1̄, . . . , k̄) = (1j1 , . . . , (dj1 )j1 , . . . , 1jl , . . . , (djl )jl ).
In next subsections an example of supersymmetric solutions will be considered. In counting the
fractional number of supersymmetries the following (2−k -splitting) theorem is used.
Theorem [246]. Let V be a vector space over K = R, C; V 6= {0}. Let Γ[i] : V → V ,
i = 1, . . . , k, be a set of linear mappings satisfying:
i, j = 1, . . . , k. Then
X
V =⊕ Vs1 ,...,sk , (12.2044)
s1 ,...,sk =±1
where
are subspaces of V , s1 , . . . , sk = ±1. Moreover, if there exists a set of linear bijective mappings
A[i] : V → V , i = 1, . . . , k, satisfying
i, j = 1, . . . , k, then all subspaces Vs1 ,...,sk are mutually isomorphic and for finite-dimensional V
s1 , . . . , sk = ±1.
CBPF-MO-003/02 276
M0 × M1 × M2 . (12.2049)
i = 1, 2.
It may be verified using relations from Subsections 7.7.3 and 7.7.4 and formulas (12.2052) and
(12.2053) that the SUSY eqs. (12.2036) are satisfied identically if
where
Γ = Γ̂11 Γ̂21 Γ̂31 . (12.2056)
Here Γ is real-valued matrix satisfying Γ2 = 1, where 1 is unit 32 × 32-matrix. Let A = Γ̂10 . The
pair Γ = Γ[1] , A = A[1] satisfies the conditions of the Theorem and hence for ε1 ∈ R32 the dimension
of the subspace Vc of solutions to eqs. (12.2055) is 16. For ε1 ∈ G321 (see Remark) the (odd part
of) superdimension of the subsuperspace Vc from (12.2055) is also 16. This means that (at least)
N = 1/2 part of SUSY is preserved.
Non-flat g i . Here we put d2 = 0 in (12.2049), i.e. we consider the metric on M0 × M1 :
with d1 = 3, d0 = 8, and 4-form from (12.2051) where metrics g i , i = 0, 1 are Ricci-flat; g 0 has
Euclidean signature and g 1 has signature diag(−1, +1, +1).
Let us consider Γ-matrices
where Γ is defined in (12.2056), η0 (x) is 2-component Killing spinor on M0 and η1 (y) is 16-component
Killing spinor on M1 , i.e.
Dµ(0) η0 = Dm
(1)
η = 0,
1 1
(12.2061)
(0) (0) (1) (1)
with Dµ = ∂µ + 14 ωabµ Γ̂a(0) Γ̂b(0) and Dm1 = ∂m1 + 14 ωa1 b1 m1 Γ̂a(1)
1
Γ̂b(1)
1
gives us the solution to SUSY
equations.
We get from (12.2058) that
Γ = Γ̂(0) ⊗ Γ̂(1) , (12.2062)
where Γ̂(1) = Γ̂1(1)
1
Γ̂2(1)
1
Γ̂3(1)
1
. Choosing real matrices Γ̂1(1)
1
= iσ2 , Γ̂2(1)
1
= σ1 , Γ̂3(1)
1
= σ3 , (where σi
are standard Pauli matrices) we get Γ̂(1) = 12 , and due to eq. (12.2062) the relation (12.2060) is
equivalent to the following one
Γ̂(0) η(0) = cη(0) . (12.2063)
Hence the number of unbroken SUSY is (at least)
where n0 (c) is the number of (chiral) Killing spinors on M0 satisfying 12.2063 with c = signν, and
n1 is the number of Killing spinors on M1 .
M 5-brane.
Now let us consider the magnetic 5-brane solution defined on the manifold (12.2049),
and (1̄, . . . , 5̄) = (10 , . . . , (d0 )0 , 12 , . . . , (d2 )2 ). Here Γ2 = 1. Let A = Γ̂11 . The pair Γ = Γ[1] , A = A[1]
satisfies the conditions of the Theorem. Hence we obtain that N = 1/2 part of supersymmetries
is preserved.
Non-flat g i . Let d2 = 0 in (12.2065), i.e. we consider the metric on M0 × M1 :
with d1 = 6, d0 = 5, where metrics g i , i = 0, 1 are Ricci-flat, g 0 has Euclidean signature and g 1 has
the signature diag(−1, +1, +1, +1, +1, +1). The 4-form (12.2066) is modified as follows
N = n0 n1 (c)/32, (12.2076)
where n1 (c) is the number of (chiral) Killing spinors on M1 satisfying (12.2075) with c = signν,
and n0 is the number of Killing spinors on M0 . A special case of this suprsymmetric solution with
M0 = R5 and M1 = R2 × K3, was considered in [247]. In this case N = 1/4 in agreemant with
(12.2076) since n0 = 4 and n1 (c) = n[K3] = 2 (the number of chiral Killing spinors on R2 × K3 is
equal to the total number of Killing spinors on K3). We remind that K3 is 4-dimensional Ricci-flat
Kähler manifold with SU (2) holonomy group and self-dual (or anti-self-dual) curvature tensor. K3
has two Killing spinors (left or right).
M 2 ∩ M 5-branes.
Here we consider solutions with two ”orthogonally” intersecting p-branes (with p = 2, 5) (see
Subsect. 3.1.1) defined on the manifold
M0 × M1 × M2 × M3 × M4 (12.2077)
i = 1, 2, where
Γ[1] = Γ̂11 Γ̂13 Γ̂23 , Γ[2] = Γ̂1̄ Γ̂2̄ Γ̂3̄ Γ̂4̄ Γ̂5̄ , (12.2083)
Acknowledgments
This work was supported in part by the DFG grant 436 RUS 113/236/O(R), by the Russian
Ministry of Science and Technology and Russian Foundation for Basic Research grant 01-17312,
CONACYT, Mexico and CBPF/MCT.
CBPF-MO-003/02 280
References
[1] V.N. Melnikov, Multidimensional Classical and Quantum Cosmology and Gravitation.Exact Solu-
tions and Variations of Constants. CBPF-NF-051/93, Rio de Janeiro, 1993;
V.N. Melnikov. In: Cosmology and Gravitation, ed. M.Novello (Editions Frontieres, Singapore, 1994)
p. 147.
[2] V.N. Melnikov, Multidimensional Cosmology and Gravitation, CBPF-MO-002/95, Rio de Janeiro,
1995, 210 p.
V.N. Melnikov. In Cosmology and Gravitation.II ed. M. Novello (Editions Frontieres, Singapore,
1996) p. 465.
[3] V.D.Ivashchuk and V.N.Melnikov. Multidimensional cosmological and spherically symmetric solu-
tions with intersecting p-branes. In Lecture Notes in Physics, Vol. 537, ”Mathematical and Quantum
Aspects of Relativity and Cosmology Proceedings of the Second Samos Meeting on Cosmology, Ge-
ometry and Relativity held at Pythagoreon, Samos, Greece, 1998, eds: S. Cotsakis, G.W. Gibbons.,
Berlin, Springer, 2000; gr-qc/9901001.
[4] K.P. Staniukovich and V.N. Melnikov, Hydrodynamics, Fields and Constants in the Theory of Grav-
itation, (Energoatomizdat, Moscow, 1983), (in Russian).
V.N. Melnikov. Fields and Constants in the Theory of Gravitation (CBPF-MO-002/02) Rio de
Janeiro, Brazil, 2002.
[5] Ya.B. Zeldovich and I.D.Novikov, Theory of Gravitation and Evolution of Stars (Nauka, Moscow,
1971) (in Russian).
[6] C. Hull and P. Townsend, Unity of Superstring Dualities, Nucl. Phys. B 438, 109 (1995).
P. Horava and E. Witten, Nucl. Phys. B 460, 506 (1996).
[7] C.M. Hull, String dynamics at strong coupling, Nucl. Phys. B 468, 113 (1996).
[10] C. Vafa, Evidence for F-Theory, hep-th/9602022; Nucl. Phys. B 469, 403 (1996).
[13] V. de Sabbata, V.N.Melnikov and P.I.Pronin, Prog. Theor.Phys. 88, 623 (1992).
[14] V.N. Melnikov. In: Gravitational Measurements, Fundamental Metrology and Constants. Eds. V. de
Sabbata and V.N. Melnikov (Kluwer Academic Publ.) Dordtrecht, 1988, p.283.
[15] A.J. Sanders and G.T. Gillies, Rivista Nuovo Cim. 19, N2, 1 (1996).
[16] A.J. Sanders and G.T. Gillies, Grav. and Cosm. 3, N4(12), 285 (1997).
[17] A.J. Sanders and W.E. Deeds. Phys.Rev.D 46, 480 (1992).
[20] T. Kaluza, Sitzungsber. Preuss. Akad. Wiss. Berlin Phys. Math., K1 33, 966 (1921).
CBPF-MO-003/02 281
[22] V. De Sabbata and E. Schmutzer, Unified Field Theories in more than Four Dimensions, (World
Scientific, Singapore, 1982).
[24] Yu.S. Vladimirov Physical Space-Time Dimension and Unification of Interactions (University Press,
Moscow, 1987) (in Russian).
[25] Yu.S. Vladimirov, The Space-time: Explicit and Hidden Symmetries, (Nauka, Moscow, 1989) (in
Russian).
[26] P.S. Wesson and J. Ponce de Leon, Gen. Rel. Gravit. 26, 555 (1994).
[27] P. Jordan, Erweiterung der projektiven Relativitatstheorie, Ann. der Phys. 219 (1947).
[28] C. Brans and R.H. Dicke, Phys. Rev. D 124, 925 (1961).
[29] E. Cremmer, B. Julia, and J. Scherk, Phys. Lett. B76 409 (1978).
[30] A. Salam and E. Sezgin, eds., Supergravities in Diverse Dimensions, reprints in 2 vols., (World
Scientific, Singapore, 1989).
[31] M.B. Green, J.H. Schwarz and E. Witten, Superstring Theory (Cambridge University Press., Cam-
bridge, 1987).
[32] V.A. Belinskii and I.M. Khalatnikov, ZhETF, 63, 1121 (1972).
[33] P. Forgacs and Z. Horvath, Gen. Rel. Grav. 11, 205 (1979).
[36] R. Abbot, S. Barr and S. Ellis, Phys. Rev. D 30, 720 (1984).
[37] V.A. Rubakov and M.E. Shaposhnikov, Phys. Lett. B 125 136 (1983).
[40] S. Ranjbar-Daemi, A. Salam and J. Strathdee, Phys. Lett. B 135, 388 (1984).
[43] M. Gleiser, S. Rajpoot and J.G. Taylor, Ann. Phys. (NY) 160, 299 (1985).
[44] U. Bleyer and D.-E. Liebscher, in Proc. III Sem. Quantum Gravity ed. M.A.Markov, V.A. Berezin
and V.P. Frolov (Singapore, World Scientific, 1985) p. 662.
[45] U. Bleyer and D.-E. Liebscher, Gen. Rel. Gravit. 17, 989 (1985).
[46] M. Demianski, Z. Golda, M. Heller and M. Szydlowski, Class. Quantum Grav. 3, 1190 (1986).
[49] G.W. Gibbons and K. Maeda, Nucl. Phys. B 298, 741 (1988).
[51] G.W. Gibbons and D.L. Wiltshire, Nucl. Phys. B 287, 717 (1987).
[52] V.D. Ivashchuk and V.N. Melnikov, Nuovo Cimento B 102, 131 (1988).
[53] K.A. Bronnikov, V.D. Ivashchuk and V.N. Melnikov, Nuovo Cimento B 102, 209 (1988).
[54] V.D. Ivashchuk and V.N. Melnikov, Phys. Lett. A 135, 465 (1989).
[55] V.D. Ivashchuk, V.N. Melnikov and A.I. Zhuk, Nuovo Cimento B 104, 575 (1989).
[56] V.A. Berezin, G. Domenech, M.L. Levinas, C.O. Lousto and N.D. Umerez, Gen. Relativ. Grav. 21,
1177 (1989).
[57] V.D. Ivashchuk and V.N. Melnikov, Chinese Phys. Lett. 7, 97 (1990).
[59] S.B. Fadeev, V.D. Ivashchuk and V.N. Melnikov, Variations of Constants and Exact Solutions in
Multidimensonal Gravity, In: Gravitation and Modern Cosmology, Plenum, N.-Y., 1991, p. 37-49.
[60] U. Bleyer, D.-E. Liebscher and A.G. Polnarev, Class. Quant. Grav. 8, 477 (1991).
[65] A.I. Zhuk, Sov. Journ. Nucl. Phys. 56, 223 (1993).
[66] C. W. Misner, In Magic without Magic: John Archibald Wheeler, ed. J. R. Klauder (Freeman, San
Francisko, 1972) p. 441.
[68] S.W. Hawking and D.N. Page, Phys. Rev. D 42, 2655 (1990).
[69] H. Liu, P.S. Wesson and J. Ponce de Leon, J. Math. Phys. 34 (1993), 4070.
[71] V.D. Ivashchuk and V.N. Melnikov, Teor. Mat. Fiz. 98 (1994), 312 (in Russian).
[72] V.D. Ivashchuk and V.N. Melnikov, Multidimensional cosmology with m-component perfect fluid,
gr-qc/ 9403063; Int. J. Mod. Phys. D 3 (1994), 795.
[73] V.R. Gavrilov, V.D. Ivashchuk and V.N. Melnikov, Integrable pseudo-euclidean Toda-like systems in
multidimensional cosmology with multicomponent perfect fluid, J. Math. Phys 36, 5829 (1995).
[74] U. Bleyer and A. Zhuk, On multidimensional cosmological models with static internal spaces, Class.
and Quantum Grav. 12, 89 (1995).
CBPF-MO-003/02 283
[75] U. Bleyer and A. Zhuk, Multidimensional integrable cosmological models with negative external
curvature, Gravitation and Cosmology, 2 106 (1995).
[76] U. Bleyer and A. Zhuk, Multidimensional integrable cosmological models with positive external space
curvature, Gravitation and Cosmology 1, 37 (1995).
[77] U. Bleyer and A. Zhuk, Kasner-like, inflationary and steady-state solutions in multidimensional
cosmology, Astron. Nachrichten, 317, 161 (1996).
[79] J.D. Barrow and J. Stein-Schabes, Phys. Rev. D 32, 1595 (1985).
[81] J. Demaret, J.-L. Hanquin, M. Henneaux, P. Spindel and A. Taormina, Phys. Lett. 175B 129 (1986).
[87] V.D. Ivashchuk and V.N. Melnikov, Billiard representation for multidimensional cosmology with
multicomponent perfect fluid near the singularity, Class. Quantum Grav. 12, 809 (1995).
[88] A.A. Kirillov and V.N. Melnikov, On Properties of Metrics Inhomogeneouties in the Vicinity of a
Singularity in K-K Cosmological Models, Astron. Astrophys. Trans., 10, 101 (1996).
[92] C. Angelantonj, L. Amendola, M. Litterio and F. Occhionero, String cosmology and inflation, Phys.
Rev. D 51, 1607 (1995).
[94] A.I. Legkii, in Probl. of Grav. Theory and Elem. Particles (Atomizdat, Moscow) 10, 149 (1979) (in
Russian).
[95] D.J. Gross and M.J. Perry, Nucl. Phys. B 226, 29 (1993).
[97] K.A. Bronnikov, V.D. Ivashchuk in Abstr. VIII Soviet Grav. Conf (Erevan, EGU, 1988) p. 156.
[98] S.B. Fadeev, V.D. Ivashchuk and V.N. Melnikov, Phys. Lett. A 161, 98 (1991).
[99] S.B. Fadeev, V.D. Ivashchuk and V.N. Melnikov, On Charged Black Hole in Multidimensional Theory
with Ricci-flat Internal Spaces, Chinese Phys. Lett. 8, 439-441 (1991).
CBPF-MO-003/02 284
[100] K.A. Bronnikov and V.N. Melnikov, Annals of Physics (N.Y.) 239, 40 (1995).
[101] U. Bleyer and V.D. Ivashchuk, Mass bounds for Multidimensional Charged Dilatonic Black Holes,
Phys. Lett. B 332, 292-296 (1994).
[102] V.D. Ivashchuk and V.N. Melnikov, Multi-temporal Generalization of the Tangherlini Solution, Class.
Quantum Grav., 11, 1793-1805 (1994).
[103] V.D. Ivashchuk and V.N. Melnikov, Multidimensional Classical and Quantum Cosmology with Per-
fect Fluid, Gravitation and Cosmology 1, No 2, 133-148 (1995).
[104] V.D. Ivashchuk and V.N. Melnikov, Gravitation and Cosmology 1, No 3, 204 (1995).
[105] V.D. Ivashchuk and V.N. Melnikov, Extremal Dilatonic Black Holes in String-like Model with Cos-
mological Term, Phys. Lett. B 384, 58 (1996).
[109] A.A. Kirillov, Pis’ma ZhETF 55, 541 (1992); Int. Journ. Mod. Phys. D 3 (1994) 1.
[110] E.I. Guendelman and A.B. Kaganovich, Phys. Lett. B 301, 15 (1993).
[112] U. Bleyer, V.D. Ivashchuk, V.N. Melnikov and A.I. Zhuk, Multidimensional classical and quantum
wormholes in models with cosmological constant. gr-qc/9405020; Nucl. Phys. B 429, 117 (1994).
[113] V.R. Gavrilov, U. Kasper, V.N. Melnikov and M. Rainer, Toda Chains with Type Am Lie Algebra for
Multidimensional m-component Perfect Fluid Cosmology, Preprint Math-97/ Univ. Potsdam, 1997.
[114] V.R. Gavrilov, V.D. Ivashchuk, and V.N. Melnikov, Class. Quant. Grav. 13, 3039 (1996).
[115] V.R. Gavrilov and V.N. Melnikov, Theor. Math. Phys 114, N3, 454 (1998).
[116] V.R. Gavrilov, V.N. Melnikov and R. Triay, Exact Solutions in Multidimensional Cosmology with
Shear and Bulk Viscosity, Class. Quant. Grav. 14, 2203 (1997).
V.R. Gavrilov, V.N. Melnikov and M. Novello, Exact Solutions in Multidimensional Cosmology with
Bulk Viscosity, Gravitation and Cosmology 1, No 2, 149 (1995).
V.R. Gavrilov, V.N. Melnikov and M. Novello, Bulk Viscosity and Entropy Production in Multidi-
mensional Integrable Cosmology Gravitation and Cosmology 2, No 4(8), 325 (1996).
[118] V.D. Ivashchuk and V.N. Melnikov, Multidimensional Gravity with Einstein Internal spaces, hep-
th/9612054; Gravitation and Cosmology 2, No 3 (7), 177 (1996).
[119] K.A. Bronnikov and J.C. Fabris, Gravitation and Cosmology 2, No 4 (8), (1996).
[120] M.J. Duff, R.R. Khuri and J.X. Lu, Phys. Rep. 259, 213 (1995).
[122] G.W. Gibbons, G.T. Horowitz and P.K. Townsend, Class. Quant. Grav. 12, 297 (1995); hep-
th/9410073.
CBPF-MO-003/02 285
[123] A. Dabholkar, G. Gibbons, J.A. Harvey, and F. Ruiz Ruiz, Nucl. Phys. B 340, 33 (1990).
[124] C.G. Callan, J.A. Harvey and A. Strominger, Nucl. Phys. 359 (1991) 611; Nucl. Phys. B 367, 60
(1991).
[125] M.J. Duff and K.S. Stelle, Phys. Lett. B 253, 113 (1991).
[126] G.T. Horowitz and A. Strominger, Nucl. Phys. B 360, 197 (1991).
[127] R. Güven, Phys. Lett. B 276, 49 (1992); Phys. Lett. B 212, 277 (1988).
[128] R. Kallosh, A. Linde, T. Ortin, A. Peet and A. van Proeyen, Phys. Rev. D 46, 5278 (1992).
[129] H. Lü, C.N. Pope, E. Sezgin and K. Stelle, Nucl. Phys. B 456, 669 (1995).
[130] A.A. Tseytlin, Mod. Phys. Lett. A11, 689 (1996); hep-th/9601177.
[131] G. Papadopoulos and P.K. Townsend, Phys. Lett. B 380, 273 (1996).
[132] A.A. Tseytlin, Harmonic Superpositions of M-branes, hep-th/9604035; Nucl. Phys. B 475, 149
(1996).
[133] J.P. Gauntlett, D.A. Kastor, and J. Traschen, Overlapping Branes in M-Theory, hep-th/9604179;
Nucl. Phys. B 478, 544 (1996).
[134] N. Khvengia, Z. Khvengia, H. Lü, C.N. Pope, Intersecting M-Branes and Bound States, hep-
th/9605082.
[135] H. Lü, C.N. Pope, and K.W. Xu, Liouville and Toda Solitons in M-Theory, hep-th/9604058.
[137] I.R. Klebanov and A.A. Tseytlin, Intersecting M -branes as Four-Dimensional Black Holes, Preprint
PUPT-1616, Imperial/TP/95-96/41, hep-th/9604166; Nucl. Phys. B 475, 164 (1996).
[138] H. Lü, C.N. Pope, and K.S.Stelle, Vertical Versus Diagonal Reduction for p-Branes, Preprint hep-
th/9605082.
[139] E. Bergshoeff, R. Kallosh and T. Ortin, Stationary Axion/Dilaton Solutions and Supersymmetry,
Preprint hep-th/9605059; Nucl. Phys. B 478, 156 (1996).
[140] G. Clément and D.V. Gal’tsov, Stationary BPS solutions to dilaton-axion gravity Preprint GCR-
96/07/02 DTP-MSU/96-11, hep-th/9607043.
[142] I.Ya. Aref’eva and A.I. Volovich, Composite p-branes in Diverse Dimensions, Preprint SMI-19-96,
hep-th/9611026; Class. Quantum Grav. 14 (11), 2990 (1997).
[143] V.D. Ivashchuk and V.N. Melnikov, Intersecting p-Brane Solutions in Multidimensional Gravity and
M-Theory, hep-th/9612089; Gravitation and Cosmology 2, No 4, 297 (1996).
[144] V.D. Ivashchuk and V.N. Melnikov, Phys. Lett. B 403, 23 (1997).
[145] I.Ya. Aref’eva, K. Viswanathan, A.I. Volovich and I.V. Volovich, p-Brane Solutions in Diverse Di-
mensions, hep-th/9701092.
[146] N. Khvengia, Z. Khvengia, H. Lănd C.N. Pope, Toward Field Theory of F-Theory, hep-th/9703012.
Class. Quant. Grav., 15, 759 (1998).
CBPF-MO-003/02 286
[147] V.D. Ivashchuk and V.N. Melnikov, Sigma-model for the Generalized Composite p-branes, hep-
th/9705036; Class. Quantum Grav. 14, 3001-30029 (1997); Corrigenda ibid. 15 (12), 3941 (1998).
[148] V.D. Ivashchuk, M. Rainer and V.N. Melnikov, Multidimensional Sigma-Models with Composite
Electric p-branes, gr-qc/9705005; Gravit. and Cosm. 4, No 1(13), 73-82 (1998).
[149] E. Bergshoeff, M. de Roo, E. Eyras, B. Janssen and J.P. van der Schaar, hep-th/9612095.
[150] I.Ya. Aref’eva and O.A. Rytchkov, Incidence Matrix Description of Intersecting p-brane Solutions,
hep-th/9612236.
[151] R. Argurio, F. Englert and L. Hourant, Intersection Rules for p-branes, hep-th/9701042.
[152] I.Ya. Aref’eva M.G. Ivanov and O.A. Rytchkov, Properties of Intersecting p-branes in Various Di-
mensions, hep-th/9702077.
[153] I.Ya. Aref’eva, M.G. Ivanov and I.V. Volovich, Non-Extremal Intersecting p-Branes in Various Di-
mensions, hep-th/9702079; Phys. Lett. B 406, 44 (1997).
[155] K.A. Bronnikov, V.D. Ivashchuk and V.N. Melnikov, The Reissner-Nordström Problem for Inter-
secting Electric and Magnetic p-Branes, gr-qc/9710054; Grav. and Cosmol. 3, No 3 (11), 203-212
(1997).
[156] K.A. Bronnikov, U. Kasper and M. Rainer, Intersecting Electric and Magnetic p-Branes: Spherically
Symmetric Solutions, gr-qc/9708058, GRG, 31, 1681 (1999).
[157] K.A. Bronnikov, M.A. Grebeniuk, V.D. Ivashchuk and V.N. Melnikov, Integrable Multidimensional
Cosmology for Intersecting p-branes, Grav. and Cosmol. 3, No 2(10), 105-112 (1997).
[158] M.A. Grebeniuk, V.D. Ivashchuk and V.N. Melnikov, Integrable Multidimensional Quantum Cos-
mology for Intersecting p-Branes, Grav. and Cosmol. 3, No 3 (11), 243-249 (1997), gr-qc/9708031.
[159] H. Lü, J. Maharana, S. Mukherji and C.N. Pope, Cosmological Solutions, p-branes and the Wheeler
De Witt Equation, hep-th/9707182.
[160] H. Lü, S. Mukherji, C.N. Pope and K.-W. Xu, Cosmological Solutions in String Theories, hep-
th/9610107.
[162] V.N. Melnikov, Multidimensional Gravity and Cosmology: Exact Solutions, Proc. 8 M.Grossman
Meeting, Jerusalem, June 1997, World Scientific, Singapore, gr-qc/9801037.
[163] V.D. Ivashchuk and V.N. Melnikov, Madjumdar-Papapetrou Type Solutions in Sigma-model and
Intersecting p-branes, Class. Quantum Grav. 16, 849 (1999); hep-th/9802121.
[164] V.D. Ivashchuk and V.N. Melnikov, Multidimensional Ouantum Cosmology with Intersecting p-
branes, Hadronic J. 21, 319 (1998).
[165] M.A. Grebeniuk, V.D. Ivashchuk and V.N. Melnikov, Multidimensional Cosmology for Intersecting p-
branes with Static Internal Spaces, gr-qc/9804042; Gravitation and Cosmology, 4, No 2(14), 145-150
(1998).
[167] N.M. Bocharova, K.A. Bronnikov and V.N. Melnikov, Vestnik MGU (Moscow Univ.), 6, 706 (1970)(in
Russian) - first MP-type solution with conformal scalar field;
K.A. Bronnikov, Acta Phys. Polonica , B4, 251 (1973);
K.A. Bronnikov and V.N. Melnikov, in Problems of Theory of Gravitation and Elementary Particles
, 5, 80 (1974) (in Russian) - first MP-type solution with conformal scalar and electromagnetic fields.
[169] G.W. Gibbons and S.W. Hawking, Phys. Rev. D 15, 2752 (1977).
[170] V.D. Ivashchuk and V.N. Melnikov, Class. and Quant. Grav. 11, 1793 (1994).
[171] V.D.Ivashchuk and V.N.Melnikov, Multitemporal Generalization of Schwarzschild Solution, Izv. Vu-
zov, Fizika, No 6 (1994) 111-112 (in Russian).
[172] V.D. Ivashchuk and V.N. Melnikov, Multitemporal Generalization of the Schwarzschild Solution. Int.
J. Mod. Phys. D 4, No 2 (1995) 167-173.
[173] S. Hewson and M. Perry, The Twelve-Dimensional Super (2+2)-Brane , Preprint hep-th/9612008.
[174] H. Nishino, Supergravity in 10+2 Dimensions as Consistent Background for Superstring, Preprint
UMDEPP 97-101, hep-th/9703214.
[178] N. Ohta and T. Shimizu, Non-extreme Black Holes from Intersecting M-branes, hep-th/9701095.
[179] V.D. Ivashchuk and V.N. Melnikov, Multidimensional Classical and Quantum Cosmology with In-
tersecting p-branes, hep-th/9708157; J. Math. Phys., 39, 2866-2889 (1998).
[183] V.V. Nikulin, On the Classification of Hyperbolic Root Systems of the Rank Three, Part I, alg-
geom/9711032.
[185] V.D. Ivashchuk and V.N. Melnikov, On Singular Solutions in Multidimensional Gravity, hep-
th/9612089; Gravitation and Cosmology 1, No 3, 204 (1996).
[186] V.G. Kac, Infinite-dimensional Lie Algebras (Cambridge University Press, Cambridge, 1990).
[187] J. Fuchs and C. Schweigert, Symmetries, Lie algebras and Representations. A graduate course for
physicists (Cambridge University Press, Cambridge, 1997).
[188] M. Adler and P. van Moerbeke, Commun. Math. Phys. 83, 83 (1982).
[191] M.A. Olshanetsky and A.M. Perelomov, Invent. Math., 54, 261 (1979).
[192] I.M. Krichever, Usp. Mat. Nauk 33 (4), 215 (1978). [in Russian].
[193] J.D. Edelstein, L. Tataru and R. Tatar, Rules for Localized Overlappings and Intersections of p-
branes, hep-th/9801049.
[194] J.M. Izquiero, N.D. Lambert, G. Papadopoulos and P.K. Townsend, Nucl. Phys. 460, 560 (1996);
hep-th/9508177.
[195] J.G. Russo and A.A. Tseytlin, Nucl. Phys. B 490, 121 (1997); hep-th/9611047.
[196] N. Ohta and J.-G. Zhou, Towards the Classification of Non-Marginal Bound States of M-branes and
Their Construction Rules, hep-th/9706153.
[197] G. Neugebauer and D. Kramer, Ann. der Physik (Leipzig) 24, 62 (1969).
[198] D. Kramer, H. Stephani, M. MacCallum, and E. Herlt, Exact Solutions of the Einstein Field Equa-
tions, CUP, Cambridge, 1980;
[199] G. Clément, Gen. Rel. and Grav. 18, 861 (1986); Phys. Lett. A 118, 11 (1986).
[200] P.G.O. Freund and M.A. Rubin, Dynamics of Dimensional Reduction, Phys. Lett. B 97, 233 (1980).
[201] M.J. Duff, B.E.W. Nilsson and C.N. Pope, Kaluza Klein Supergravity, Phys. Rep. 130, 1 (1986).
[202] G.W. Gibbons and P.K. Townsend, Vacuum Interpolation in Supergravity via Super p-branes, Phys.
Rev. Lett. 71 3754 (1993).
[203] L. Castellani, A. Ceresole, R. D’Auria, S. Ferrara, P. Fré and M. Trigiante, G/H M-branes and
AdSp+2 Geometries, hep-th/9803039.
[205] J. Maldacena, The Large N Limit of Superconformal Field Theories and Supergravity, hep-
th/9711200.
[206] S. Ferrara and C. Fronsdal, Conformal Maxwell Theory as a Singleton Field Theory on AdS5 , IIB
Three-Branes and Duality, hep-th/971223.
[207] S.S. Gubser, I.R. Klebanov and A.M. Polyakov, Gauge Theory Correlators from Non-Critical String
Theory, hep-th/9802109.
[209] A. Zhuk, Generalized deSitter Solution in Multidimensional Cosmology with Static Internal Spaces,
Astron. Nachrichten 316 269 (1995).
[210] H.J. Boonstra, B. Peeters and K. Skenderis, Brane Intersections, Anti-de Sitter Spacetimes and Dual
Superconformal Theories, hep-th/9803231.
[211] U. Kasper and A. Zhuk, Integrable Multicomponent Perfect Fluid Multidimensional Cosmology.1,
GRG, 28 1269 (1996).
[212] U. Günther and A.Zhuk, Gravitational excitons from extra dimensions. Phys. Rev. D 56, 6391 (1997).
[213] S. Helgason, Differential Geometry and Symmetric Spaces, New York: Academic Press (1962).
CBPF-MO-003/02 289
[214] K.A. Bronnikov, Block-orthogonal Brane systems, Black Holes and Wormholes, hep-th/9710207;
Gravitation and Cosmology 4, No 1 (13), 49 (1998).
[215] D.V. Gal’tsov and O.A. Rytchkov, Generating Branes via Sigma models, hep-th/9801180.
[216] P. Fre, Solvable Lie Algebras, BPS Black Holes and Supergravity Gaugings, hep-th/9802045.
[217] M.A. Grebeniuk and V.D. Ivashchuk, Sigma-model Solutions and Intersecting p-branes Related to
Lie Algebras, Phys. Lett. B 442, 125 (1998); hep-th/9805113.
[219] V.D. Ivashchuk, S.-W. Kim and V.N. Melnikov, Hyperbolic Kac-Moody Algebra from Intersecting
p-branes, J. Math. Phys. 40, 4072 (1999); hep-th/9803006.
[220] V.D. Ivashchuk, Composite p-branes on Product of Einstein spaces, Phys. Lett., B 434, No 1, 28-35,
(1998); hep-th/9704113.
[221] M.A. Grebeniuk, V.D. Ivashchuk and V.N. Melnikov. P -brane Multi- dimensional Cosmology with
Spontaneous Compactification of Internal Spaces. Gravitation and Cosmology, 5, Supplement, 1-8
(1999).
[222] V.R. Gavrilov and V.N. Melnikov, Toda Chains with Type Am Lie Algebra for Multidimensional
Classical Cosmology with Intersecting p-branes, In : Proceedings of the International seminar ”Curent
topics in mathematical cosmology”, (Potsdam, Germany , 30 March - 4 April 1998), Eds. M. Rainer
and H.-J. Schmidt, World Scientific, 1998, p. 310; hep-th/9807004.
[223] S. Cotsakis, V.D. Ivashchuk and V.N. Melnikov, P-branes Black Holes and Post-Newtonian Approx-
imation, Grav. and Cosmol. 5, No 1 (17), (1999); gr-qc/9902148.
[224] V.R. Gavrilov and V.N. Melnikov, Toda Chains Associated with Lie Algebras Am in Multidimensional
Gravitation and Cosmology with Intersecting p-branes, Theor. Math. Phys. (2000) 123 No 3, 374-394
(in Russian).
[225] V.D.Ivashchuk and S.-W. Kim. Solutions with intersecting p-branes related to Toda chains, J.
Math.Phys. 2000, 41 (1).
[226] M. Toda, Progr. Theor. Phys. 45, 174 (1970); Theory of Nonlinear Lattices (Springer-Verlag, Berlin,
1981).
[228] K.A. Bronnikov, Gravitating Brane Systems: Some General Theorems, gr-qc/9806102; J. Math.
Phys., 40, 924 (1999).
[229] V.D.Ivashchuk and V.N.Melnikov. Billiard Representation for Multi- dimensional Cosmology with
Intersecting p-branes near the Singularity. J. Math. Phys., 41, No 8, (2000); hep-th/9904077.
[230] V.D.Ivashchuk and V.N.Melnikov. Billiard representation for multidimensional cosmology with p-
branes near the singularity. In Advanced Series in Astrophysica and Cosmology-Vol. 10. ”The Chaotic
Universe”, Proc. of the Second ICRA Network Workshop, Eds. V.G. Gurzadyan and R. Ruffini, 1999,
World Scientific, Singapore, p. 509-524.
[231] P. Dobiash and D. Maison, Gen. Rel. Grav. 14, 231 (1982).
[232] D.J. Gross and M.J. Perry, Nucl. Phys. B 226, 29 (1983).
CBPF-MO-003/02 290
[234] G. Gibbons and D. Wiltshire, Ann. Phys. 167, 201 (1986); Erratum: ibid 176, 393 (1987).
[235] C.-M. Chen, D. V. Gal’tsov, K. Maeda and S. Sharakin, Phys. Lett. B 453, 7 (1999).
[236] C.-M. Chen, D. V. Gal’tsov, and S. Sharakin, Einstein Gravity — Supergravity Correspondence,
hep-th/9912127.
[237] V.D.Ivashchuk and V.N.Melnikov. P-brane black Holes for General Intersections. Gravitation and
Cosmology, , 5, No 4 (20) 313-318 (1999); gr-qc/0002085.
[238] V.D.Ivashchuk and V.N.Melnikov. Black hole p-brane solutions for general intersection rules. Gravi-
tation and Cosmology, 2000, 6, No 1 (21) 27-40; hep-th/9910041.
[239] V.D.Ivashchuk and V.N.Melnikov. Toda p-brane black holes and polynomials related to Lie algebras.
Class. and Quantum Gravity 17 2073-2092 (2000); math-ph/0002048.
[240] A. Rogers, J. Math. Phys., A Global Theory of Supermanifolds, 22, No 5, (1981) 939-945; J. Math.
Phys., Super Lie Groups: Global Topology and Local Structure, 21, No 6 (1980) 724-731; J. Math.
Phys., Consistent Superspace Integration, 26, No 3, (1985) 385-392.
[243] V.D. Ivashchuk, Tensor Banach Algebras of Projective Type I. Teor. Mat. Fiz., 91, No 1, (1992)
17-29 [in Russian].
[244] V.D. Ivashchuk, Tensor Banach Algebras of Projective Type II. l1 - case, Teor. Mat. Fiz., 91, No 2
(1992) 192-206 [in Russian].
[247] A. Kaya, A Note on a Relation between Killing Spinor and Einstein Equations, hep-th/9902010;
Phys. Lett. B 458, 267 (1999).
[248] V.D. Ivashchuk, V.S. Manko and V.N. Melnikov, Post-Newtonian parameters for general black hole
and spherically symmetric p-brane solutions, Grav. Cosmol. 6, No 3 (23), 219 (2000); gr-qc/0101044.
[249] V.D. Ivashchuk, M. Kenmoku and V.N. Melnikov, On quantum analogues of p-brane black hole,
Grav. Cosmol. 6, No 3 (23), 225 (2000); gr-qc/0101043.
[253] U. Bleyer, D.-E. Liebscher, H.-J. Schmidt and A. I. Zhuk, Wissenschaftliche Zeitschrift, 39 (1990)
20.
[255] N.N. Bogoljubov and V.D. Shirkov, ”Introduction to the Theory of Quantized fields”, Nauka, Moscow,
1984 (in Russian).
[256] N. Birrell and P. Davies, ”Quantized Fields in Curved Space-Time”, Cambridge University Press,
1980.
[257] A.A. Grib, S.G. Mamaev and V.M. Mostepanenko, ”Vacuum Quantum Effects in Strong Fields”,
Friedmann Laboratory Publishing, St. Petersburg, 1994.
[260] V.D. Ivashchuk, Regularization by ²-metric.I, Izvest. Akad. Nauk Mold. SSR, Ser. fiz.-tekhn. i mat.
nauk, 3, (1987) 8-17 (in Russian).
[261] V.D. Ivashchuk, Regularization by ²-metric.II. The limit ² = +0, Izvest. Akad. Nauk Mold. SSR, Ser.
fiz.-tekhn. i mat. nauk, 1, (1988) 10-20 (in Russian).
[264] E. Elizalde, S.D. Odintsov and A. Romero, Class. Quantum Grav. 11 (1994), L61.
[265] V.D. Ivashchuk, Wick Rotation, Regularization of Propagators by a Complex Metric and Multidi-
mensional Cosmology, Grav. Cosmol., 3, No 1 (9), 8-16 (1997).
[270] V.A. Belinskii, E.M. Lifshitz and I.M. Khalatnikov, Usp. Fiz. Nauk 102, 463 (1970) [in Russian];
Adv. Phys. 31, 639 (1982).
[271] C.W. Misner, The Mixmaster cosmological metrics, preprint UMCP PP94-162; gr-qc/9405068.
[272] D.V. Anosov, Geodesic flows on manifolds of constant curvature, Trudi of Steklov Math. Inst.
(Moscow) 1967 [in Russian]
[273] I.P. Kornfeld, Ya.G. Sinai and S.V. Fomin, Ergodic theory (Moscow, Nauka) 1980 [in Russian].
[275] V.G. Boltyansky and I.Z. Gohberg, Theorems and Problems of Kombinatorial Geometry (Moscow,
Nauka) 1965 [in Russian].
[276] L. Fejes Toth, Lagerungen in der Ebene auf der Kugel und Raum (Berlin, Springer) 1953.
[279] C. Misner, K. Thorne and J. Wheeler, Gravitation (San Francisco, Freeman & Co.) 1972
CBPF-MO-003/02 292
[280] J. Pullin , Time and Chaos in General Relativity preprint Syracuz. Univ. Su-GP-91/1-4.; 1991 in
Relativity and Gravitation: Classical and Quantum, Proc. of SILARG VII Cocoyos Mexico 1990
(Singapore, World Scientific) ed. D’Olivo J C et al.
[282] V.D. Ivashchuk and V.N. Melnikov, Billiard Representation for Multidimensional Quantum Cosmol-
ogy near the Singularity, preprint RGA-CSVR-014/94; gr-qc/ 9411012.
[283] I.Ya. Aref’eva, P.B. Medvedev, O.A. Rytchkov and I.V. Volovich, Chaos in M(atrix) Theory, hep-
th/9710032.
[284] R.C. Myers and M.J. Perry, Ann. of Phys., 172 (1986) 304.
[285] U. Bleyer, K. A. Bronnikov, V. N. Melnikov and S. B. Fadeev, On black hole stability in multidi-
mensional gravity, AIP preprint (Potsdam) 94-01, 1994.
[288] V.D. Ivashchuk, V.N. Melnikov and S.B. Fadeev, Spherically- symmetric solution of Einstein-Maxwell
equations with with Ricci-flat Internal Spaces, Izv. Vuzov, Fizika, No 10 (1994), 113-114 (in Russian).
[289] D. Garfinkle, G. Horowitz and A. Strominger, Phys. Rev., D43 (1991) 3140; D45 (1992) 3888(E).
[291] G.W. Gibbons and C.G. Wells, Anti-Gravity Bounds and the Ricci Tensor, DAMTP preprint R93/25,
1993; Commun. Math. Phys..
[292] N.S. Kalitzin, Wissenschaftliche Zeitschrift der Humboldt- Universitat zu Berlin, Jg. VII Nr 2, 207
(1957/58).
[293] N.S. Kalitzin, Izv. Bolgar. Ak. Nauk, Fiz. 7 (1959) 219.
[296] J.H. Horne and G.T. Horowitz, Phys. Rev. D 48 (1993) R5457.
[298] K. Shiraishi, J. Math. Phys. 34 (1993) 1480; Nucl. Phys. B 402 (1993) 399.
[299] S.J. Poletti and D.L. Wiltshire, Phys. Rev. D 50 (1994) 7260.
[303] C.G. Callan, R.C. Myers and M.J. Perry, Nucl. Phys. B 311 (1988) 673.
[304] K.A. Bronnikov, V.D. Ivashchuk and V.N. Melnikov in: Problems of gravitation, Plenary reports.
7th Soviet Conf. on Gravitation (Erevan: ERGU, 1980), p. 70.
CBPF-MO-003/02 293
[305] V. N. Melnikov, in: Itogi Nauki i Tekhniki. Classical Field Theory and Theory of Gravity. Vol 1.
Gravitation and Cosmology. (Moscow: VINITI, 1991) p. 49 [in Russian].
[309] T. Kugo and P.K. Townsend, Nucl. Phys. B 266 (1983) 440.
[310] van Nieuwenhuizen P 1983 in: Relativity, groups and topology, ed., DeWitt and Stora (North-
Holland) Amsterdam
[311] I.Ya. Aref’eva and I.V. Volovich, Phys. Lett. B 164 (1985) 287.
[312] M.P. Blencowe and M.J. Duff, Nucl. Phys. B 310 (1988) 387.
[313] C.M. Hull and N.P. Warner, Class. Quantum. Grav. 5 (1988) 1517.
[315] A.V. Kamenev, Probl. of Grav. Theory and Relat. Theory (Moscow:UDN, 1986) p. 20 [in Russian].
[316] V.N. Melnikov, Gravity as a Key Problem of the Millennium. Proc. 2000 NASA/JPL Conference
on Fundamental Physics in Microgravity, CD-version, NASA Document D-21522, 2001, p. 4.1-4.17,
(Solvang, CA, USA); gr-qc/0007067.
V.N. Melnikov, Gravitation and Cosmology, 6, N2(22) (2000) 81.
[317] A. Miyazaki, Time-Variation of the Gravitational Constant and the Machian Solution in the Brans-
Dicke Theory, gr-qc/0102003.
[323] Staff of the Space Department, John Hopkins University, Applied Lab., and Staff of the Guidance and
Control Lab., Stanford University, 1974. A satellite freed of all but gravitational forces: TRIAD-1,
J. Spacecraft and Rockets, v.11, 637-644.
[324] Sanders A.J. et al. Project SEE: an international effort to develop a space-based mission for precise
measurements of gravitation. Class. Quant. Grav. 17. (2000), 2331-2346.
A. Sanders, V.N. Melnikov et al., Class. Quant. Grav. 17 (2000) 2331.
V.N. Melnikov and A. Sanders, Ciencia Ergo Sum, 8 (2001) 357.
[325] Braginsky V.B., Manukin A.B. Measurements of small forces in physical experiments. M., ”Nauka”,
1974. (In Russian).
CBPF-MO-003/02 294
[326] Ciufolini I., et al. Test of General Relativity and Measurement of the Lense-Thirring Effect with Two
Earth Satellites. Science, vol.279, (1998), 2100-2103.
[327] V.D. Ivashchuk and V.N. Melnikov. Class. Quant. Grav., 18, R1-R66 (2001).
[328] M.J. Duff, H. Lü, and C.N. Pope. ”The Black Branes of M-theory”, hep-th/9604052.
[329] H. Lü, and C.N. Pope. ”SL(N + 1, R) Toda Solitons in Supergravities”, hep-th/9607027.
[330] H. Lü, and C.N. Pope. ”Black p-branes and Their Vertical Dimensional Reduction”, hep-th/9609126.
[332] A.N. Leznov and M.V. Saveliev. Group-theoretical Methods for Integration of Nonlinear Dynamical
Systems. Birkhauser, Basel, 1992.
[333] S. Cotsakis, V.R. Gavrilon and V.N. Melnikov. Grav. Cosmol. 5, 67-78 (1999).
[334] V.R. Gavrilov and V.N. Melnikov, in: “Current Topics in Mathematical Cosmology”, Proc. Int. Sem.
on Mathematical Cosmology, Potsdam, April 1998; WS, Singapore, p. 310.
[338] T. Regge and J.A. Wheeler, Phys. Rev. 108, 1063 (1957).
[341] K.A. Bronnikov and Yu.N. Kireyev, Phys. Lett. A 67, 95 (1978).
[342] K.A. Bronnikov and A.V. Khodunov, Gen. Rel. & Grav. 11, 13 (1979).
[343] P. Kanti and E. Winstanly, “Do Stringy Corrections Stabilize Coloured Black Holes?”, gr-qc/9910069.
[344] K.D. Kokkotas and B.G. Schmidt, “Quasi-Normal Modes of Stars and Black Holes”, gr-qc/9909058.
[345] R. Gregory and R. Laflamme, Phys. Rev. Lett. 70, 2387 (1993); hep-th/9301052.
[346] R. Gregory and R. Laflamme, Nucl. Phys. B 428, 399 (1994); hep-th/9494071.
[347] R. Gregory and R. Laflamme, Phys. Rev. D 51, 7007 (1995); hep-th/9410050.
[349] P. Kanti, N.E. Mavromatis, J. Rizos, K. Tamvakis and E. Winstanly, Phys. Rev. D 57, 6255 (1998).
[351] F.R. Gantmacher, “Matrix Theory”, Nauka, Moscow, 1988 (in Russian).
[352] K.A. Bronnikov and V.N. Melnikov. Nucl. Phys. B584, 436-458 (2000).
[353] C.M. Will, “Theory and Experiment in Gravitational Physics”, Cambridge University Press, Cam-
bridge, 1993.
CBPF-MO-003/02 295
[355] J.K. Webb, M.T. Murphy, V.V. Flambaum, V.A. Dzuba, J.D. Barrow, C.W. Churchill, J.X.
Prochaska and A.M. Wolfe, “Further evidence for cosmological evolution of the fine structure con-
stant”, astro-ph/0012539.
[358] L. Randall and R. Sundrum, Phys. Rev. Lett. 83,, 4690 (1999); hep-th/9906064.
[359] M.D. Maia and V. Silveira, Phys. Rev. D 48, 954 (1993).
[360] I. Bars and C. Kounnas, Phys. Rev. D 56, 3664 (1997); hep-th/9703060;
H. Nishino, Phys. Lett. 428B, 85 (1998); hep-th/9703214.
[361] C.M. Hull and R.R. Khuri, Nucl. Phys. B 575, 231 (2000) hep-th/9911082.
[364] G.T. Horowitz and A. Strominger, Nucl. Phys. B 360, 197 (1991).
[365] G. Magnano and L.M. Sokolowski, Phys. Rev. D 50, 5039 (1994); gr-qc/9312008.
[366] V. Faraoni, E. Gunzig and P. Nardone, “Conformal transformations in classical gravitational theories
and in cosmology”, gr-qc/9811047; Fundamentals of Cosmic Physics 20, 121 (1999).
[367] M. Rainer and A.I. Zhuk, Gen. Rel. Grav. 32, 79 (2000); gr-qc/9808073.
[368] S. Nojiri, O. Obregon, S.D. Odintsov and V.I. Tkach, “String versus Einstein frame in AdS/CFT
induced quantum dilatonic brane-world Universe”, hep-th/0101003.
[370] T.M. Eubanks at al., “Advances in solar system tests of gravity”, preprint, available at
ftp://casa.usno.navy.mil/navnet/postscript/, file prd 15.ps (1999).
[373] T. Jacobson and G. Kang, Clas. Qu. Grav. 10, L201 (1993).
[374] K.A. Bronnikov, Izvestiya Vuzov, Fizika, 1979, No. 6, 32 (in Russian).
[375] K.A. Bronnikov and V.N. Melnikov. GRG, 33, 1549-1578 (2001).
[378] M. Gogberashvili, Phys. Lett. B 484, 124 (2000) (see also hep-ph/9812296; hep-ph/9812365; hep-
ph/9904383; hep-ph/9908347).
CBPF-MO-003/02 296
[379] P. Binétruy, C. Deffayet and D. Langlois, Nucl. Phys. B 565, 269 (2000); hep-th/9905012.
[380] N. Arkani-Hamed, S. Dimopoulos, G. Dvali and N. Kaloper, Phys. Rev. Lett. 84, 586 (2000); hep-
th/9907209.
[381] L. Randall and R. Sundrum, A large mass hierarchy from a small extra dimension, Phys. Rev. Lett.
83, 3370 (1999); hep-ph/9905221.
[382] L. Randall and R. Sundrum, An alternative to compactification, Phys. Rev. Lett. 83, 4690 (1999);
hep-th/9906064.
[383] I.I. Kogan, S. Mouslopoulos, A. Papazoglou, G.G. Ross, A three three-brane universe: New phe-
nomenology for the new millennium? Nucl. Phys. B 584, 313 (2000); hep-ph/9912552.
[384] G. Dvali, G. Gabadadze and G. Senjanović, Constraints on extra time dimensions, hep-ph/9910207.
[385] T. Shiromizu, K. Maeda and M. Sasaki, Phys. Rev. D 62, 024012 (2000).
[386] C. Csáki, J. Erlich, T.J. Hollowood and Y. Shirman Universal Aspects of Gravity Localized on Thick
Branes, Nucl. Phys. B 581, 309 (2000); hep-th/0001033.
[387] S. Nojiri and S.D. Odintsov, Phys. Lett. B 484 119 (2000); hep-th/0004097;
S. Nojiri, S.D. Odintsov and S. Zerbini, Phys. Rev. D 62, 064006 (2000), hep-th/0001192; for a
review, see S. Nojiri and S.D. Odintsov, hep-th/0105160.
[388] S.W. Hawking, T. Hertog and H.S. Reall, Phys. Rev. D 62 (2000) 043501; hep-th/0003052.
[389] R. Gregory, V.A. Rubakov and S.M. Sibiryakov, Opening up extra dimensions at ultra-large scales,
Phys. Rev. Lett. 84, 5928 (2000), hep-th/0002072.
[390] W. Mück, K.S. Viswanatan and I.V. Volovich, Geodesic and Newton’s Law in Brane Backgrounds,
Phys. Rev. D 62 105019 (2000); hep-th/0002132.
[391] I.Ya. Aref’eva, M.G. Ivanov, W. Mück, K.S. Viswanatan and I.V. Volovich, Consistent Linearized
Gravity in Brane Backgrounds, Nucl. Phys. B 590, 273 (2000); hep-th/0004114.
[392] G. Gibbons, R. Kallosh and A. Linde, Brane World Sum Rules, JHEP 0101, 022 (2001); hep-
th/0011225.
[393] A. Iglesias and Z. Kakushadze, Time-like Extra Dimensions without Tachyons or Ghosts, hep-
th/0012049.
[394] Yu.V. Shtanov, Flat vacuum branes without fine tuning, hep-ph/0108211.
[395] V. Sahni and A. Starobinsky, The case for a Positive Cosmological Lambda-term, Int. J. Mod. Phys.,
D 9, 373-444 (2000).
[396] V.D. Ivashchuk and V.N. Melnikov. Grav. Cosm. 7, N3 (27), 241-245 (2001).
[397] Zaitsev V F and Polyanin A D 1994 Discrete-Groups Methods for Integrating Equations of Nonlinear
Mechanics (Roca Raton, CRC Press-Begel House)
[398] Polyanin A D and Zaitsev V F 1995 Handbook on Exact Solutions for Ordinary Differential Equations
(Roca Raton, CRC Press)
[402] Börner G 1992 The Early Universe. Facts and Fiction (Berlin: Springer)
[403] Gavrilov V R, Ivashchuk V D, Kasper U and Melnikov V N, 1997 Gen. Relativ. Grav. 29 599
[408] Abramowitz A, Stegun I A, 1964 Handbook on Mathematical Functions ( National Bureau of stan-
dards, Applied Math. Series 55).
[409] Belinsky V A and Khalatnikov I M 1975 Soviet Phys. JETP Letters 21 99.
[413] Collins C B and Stewart J M 1971 Monthly Notices Roy. Astron. Soc. 151 419.
[414] Gavrilov V R, Melnikov V N and Triay R Exact solutions in multidimensional cosmology with bulk
and shear viscosity Preprint CPT-96/P.3396, Marseille, France.
[422] Oliveira H P and Salim J M 1988 Acta Phys. Pol. B19 649.
[429] I. Waga and A. Miceli, ”Cosmological constraints from lensing statistics and supernova on the cosmic
equation of state”, astro-ph/9811460.
CBPF-MO-003/02 298
[430] V.B. Johri, ”Constraints over cosmological constant and quintessence fields in an accelerating Uni-
verse”, astro-ph/00070079.
[431] S. Bludman and M. Roos, ”Vacuum energy: cosmological constant or quintessence?”, astro-
ph/0004328.
[432] I. Zlatev, L. Wang and P.J. Steinhardt, ”Quintessence, cosmic coincidence and the cosmological
constant”, astro-ph/9807002.
[433] R. de Ritis and A.A. Marino, ” Effective cosmological ”constant” and quintessence”, astro-ph/007128.
[434] P. Brax and J. Martin, ”High energy physics and quintessence”, astro-ph/0005449.
[435] L.S. Chimento and A.S. Jakubi, ”Enlarged Q-matter cosmology”, astro-ph/005070.
[436] V.R. Gavrilov and V.N. Melnikov. Gravitation and Cosmology, 2001, v. 7, N4 (28), p. 301-307.