Principles of Einstein-Finsler Gravity and Perspectives in Modern Cosmology

October 2, 2012 3:46 WSPC/S0218-2718 142-IJMPD 1250072
International Journal of Modern Physics D

Vol. 21, No. 9 (2012) 1250072 (40 pages)
c World Scientic Publishing Company
DOI: 10.1142/S0218271812500721
PRINCIPLES OF EINSTEINFINSLER GRAVITY
AND PERSPECTIVES IN MODERN COSMOLOGY
SERGIU I. VACARU
Science Department, University Al. I. Cuza Iasi
54 Lascar Catargi Street, Iasi, Romania 700107
sergiu.vacaru@uaic.ro
Sergiu.Vacaru@gmail.com
Received 23 December 2011
Revised 29 July 2012
Accepted 11 August 2012
Published 4 October 2012
We study the geometric and physical foundations of Finsler gravity theories with metric
compatible connections dened on tangent bundles, or (pseudo) Riemannian manifolds,
endowed with nonholonomic frame structure. Several generalizations and alternatives
to Einstein gravity are considered, including modications with broken local Lorentz
invariance. It is also shown how such theories (and general relativity) can be equivalently
re-formulated in Finsler like variables. We focus on prospects in modern cosmology
and Finsler acceleration of Universe. EinsteinFinsler gravity theories are elaborated
following almost the same principles as in the general relativity theory but extended to
Finsler metrics and connections. Finally, some examples of generic o-diagonal metrics
and generalized connections, dening anisotropic cosmological EinsteinFinsler spaces
are analyzed; certain criteria for the Finsler accelerating evolution are formulated.
Keywords: Modied gravity; cosmology; Finsler geometry.
1. Introduction
During last 30 years, the experimental data and existing methodology and phe-
nomenology of particle physics, and gravity, imposed an interpretation doctrine
that models of Finsler like spacetimes (with metrics and connections depending on
velocity/momenta) are subjected to strong experimental restrictions. Such theo-
ries were not included in the standard paradigm of modern physics (see respective
arguments in Refs. 1 and 2
a
).
Nevertheless, there are various theoretical arguments
39
that quantum grav-
ity (QG) models positively result in nonlinear dispersion relations depending on
a
These studies did not include all fundamental geometric/physical objects in Finsler geome-
try/gravity, for instance, the nonholonomic structure, nonlinear connections, N-connections, and
new types of linear connections which are adapted to N-connections, the possibility to model
(pseudo) Finsler geometric models as exact solutions in Einstein gravity, etc.
1250072-1
S. I. Vacaru
velocities/momenta. Anisotropic quasiclassical Finsler congurations originating
from QG are not obligatory restricted for some inationary cosmological models
and may have important contributions to dark energy and dark matter in universe.
This constrains us to investigate Finsler type spacetimes both in QG theories and
modern cosmology.
In a survey
10
oriented to nonexperts in Finsler geometry (but researches in par-
ticle physics and gravity) we discussed in detail and formulated well dened criteria
how the Finsler geometry methods and theories with nonholonomic distributions
b
can be elaborated following the standard paradigm of modern physics. The pur-
pose of the present work is to study the fundamental principles of EinsteinFinsler
gravity theories
c
and analyze possible further applications in modern cosmology.
Almost all classical gravitational eects are described in the framework of gen-
eral relativity (GR). Recently, it was proposed that certain exceptions in theoretical
cosmology may be related to the dark matter and dark energy problems and some
approaches were formulated for nonmetric Finlser gravity models. Here we note
that the existence of anisotropies and inhomogeneities has become a conventional
feature in cosmology physics. A number of cosmological models with Finsler met-
rics have been elaborated by now, and this number seems to grow rapidly.
1114
It
is also expected that small corrections with violations of equivalence principle and
local Lorentz invariance have to be considered in low energy limits within general
approaches to QG (see Refs. 3, 5, 79, 15 and references therein for a series of works
related to Finsler geometry).
There are two general classes of Finsler type gravity theories with very dif-
ferent implications in physics, mechanics and cosmology. The rst class originates
from Cartan works on Finsler geometry
16
(see further geometrical developments
and applications in Refs. 1722 and 10). In those works a number of geometric
and physical constructions and Finsler geometry methods were considered for both
types of metric compatible or noncompatible connections. The most related to
standard physics constructions were elaborated for the metric compatible Car-
tan and canonical distinguished connections (in brief, d-connections; see details in
Refs. 19 and 10) following geometric and physical principles which are very similar
to those used for building the GR theory.
6
In the second class of theories, there
are Finsler geometry and gravity models derived for the Berwald and Chern d-
connections which are not metric compatible (see details in Refs. 23, 13 and 14;
b
A pair (V, N), where V is a manifold and N is a nonintegrable distribution on V, is called a
nonholonomic manifold. Modeling Finsler like geometries in Einstein gravity, we have to consider
that V is a four-dimensional (pseudo) Riemannian spacetime when the LeviCivita connection
is correspondingly deformed to a linear connection adapted to a N-connection structure dened
by a nonintegrable (2+2)-splitting (i.e. nonholonomic distribution for frame elds). The boldface
symbols will be used for nonholonomic manifolds/bundles and geometric objects on such spaces,
as we discuss in details.
10,22,26
c
Such a model of gravity is constructed following the same principles as the GR theory but on tan-
gent bundles, or on (pseudo) Riemannian manifolds endowed with nonholonomic distributions, for
a (Finsler-type) metric compatible linear connection adapted to a nonlinear connection structure.
1250072-2
Principles of EinsteinFinsler Gravity
summaries of results and applications to (non)standard physical theories are given
in Part I of Refs. 22 and 10).
This paper is organized as follows: In Sec. 2, we provide physical motivations
for Finsler gravity theories and explain in brief how fundamental Finsler geometric
objects are dened of tangent bundles and in Einstein gravity. In Sec. 3, we consider
the gravitational eld equations for FinslerEinstein gravity. We briey discuss how
generic o-diagonal
d
solutions can be constructed in exact form and provide some
examples. Two classes of cosmology diagonal and o-diagonal solutions on tangent
bundles modeling Finsler acceleration of Universe are analyzed in Sec. 4. Finally,
in Sec. 5 we outline the approach and formulate conclusions. In Appendices, we
provide local formulas and examples of cosmological solutions.
2. (Pseudo) Finsler Spacetimes
In this section, the most important geometric constructions and physical motiva-
tions for Finsler gravity are summarized. We shall follow the system of notations
proposed in Refs. 10 and 22 (see details and references therein).
e
2.1. Physical motivations for Finsler gravity theories
The rst example of a Finsler metric
24
was given by Riemann,
25
who consid-
ered forth-order forms instead of quadratic line elements (see historical remarks
in Refs. 19 and 23 and, in relation to standard and nonstandard physical theories,
in Refs. 10, 22, 26, 21). In this subsection, we show that Finsler like nonlinear line
elements are generated naturally by deformations of standard Minkowski metrics
in special relativity (SR) (see also Refs. 2730, 4, 31, 11 and 5).
2.1.1. Violations of Lorentz symmetry and Finslerian Hessians
Finsler metrics can be generated if instead of the Lorentz transforms in SR there
are considered nonlinear generalizations, restrictions of symmetries and/or defor-
mations of the Minkowski metric (see examples in Refs. 32 and 11). We provide a
simple construction when anisotropic metrics are used for modeling light propaga-
tion in anisotropic media (aether) and/or for small perturbations in QG.
In SR, a Minkowski metric
ij
=diag[1, +1, +1, +1] (for i =1, 2, 3, 4) denes
a quadratic line element, ds
2
=
ij
dx
i
dx
j
= (dx
1
)
2
+ (dx
2
)
2
+ (dx
3
)
2
+ (dx
4
)
2
.
The light velocity c is contained in x
1
=ct, where t is the time-like co-
ordinate. Along a light ray x
i
(), parametrized by a real smooth parameter
0
0
, when ds
2
/d
2
=0, we can dene a null tangent vector eld
y
i
() =dx
i
/d, with d =dt/d. Under general co-ordinate transforms x
i
=x
i
(x
i
),
d
Which cannot be diagonalized by co-ordinate transforms.
e
In this work, we do not aim to provide an exhaustive bibliography (it is not possible to list and
discuss tenths of monographs and thousands of articles on Finsler geometry and generalizations
and various extensions of the GR theory.
1250072-3
S. I. Vacaru
we have
ij
g
i
j
(x
k
); the condition ds
2
/d
2
=0 holds always for propaga-
tion of light, i.e. g
i
j
y
i
y
j
=0. We can write for some classes of co-ordinate

systems c
2
=g
b
i
b
j
(x
i
)y
b
i
y
b
j
/
2
(for simplicity, omitting priming of indices and writing
i,

j, . . . =2, 3, 4).
f
In anisotropic media (and/or modeling spacetime as an aether model), we can
generalize the quadratic expression g
b
i
b
j
(x
i
)y
b
i
y
b
j
to an arbitrary nonlinear one

F
2
(y
b
j
)
subjected to the condition of homogeneity that

F(y
b
j
) =

F(y
b
j
), for any > 0.
The formula for light propagation transforms into c
2
=

F
2
(y
b
j
)/
2
. Small deforma-
tions of the Minkowski metric can be parametrized in the form

F
2
(y
b
j
) (
b
i
b
j
y
b
i
y
b
j
)
r
+
q
b
i
1
b
i
2
b
i
2r
y
b
i
1
y
b
i
2r
, for r = 1, 2, . . . and

i
1
,
i
2
, . . . ,
i
2r
= 2, 3, 4. In the approxima-
tions r = 1 and q
b
i
1
b
i
2
b
i
2r
0, we get the Minkowski (pseudo-Euclidean) spacetime
in SR. We can generalize the coecients of

F
2
by introducing additional dependen-
cies on x
i
, when

F
2
(x
i
, y
b
j
) (g
b
i
b
j
(x
k
)y
b
i
y
b
j
)
r
+q
b
i
1
b
i
2
b
i
2r
(x
k
)y
b
i
1
y
b
i
2r
, and consider
certain generalized nonlinear homogeneous relations (with F(x
i
, y
j
) = F(x
i
, y
j
),
for any > 0), when
ds
2
= F
2
(x
i
, y
j
)
(cdt)
2
+g
b
i
b
j
(x
k
)y
b
i
y
b
j
1 +
1
r
q
b
i
1
b
i
2
b
i
2r
(x
k
)y
b
i
1
y
b
i
2r
(g
b
i
b
j
(x
k
)y
b
i
y
b
j
)
r
+O(q
2
). (1)
A nonlinear element ds
2
= F
2
(x
i
, y
j
) is usually called by physicists a Finslerian
metric. In the bulk of geometric constructions in books,
19,23
the function F is
considered to be a fundamental (and/or generating) Finsler function satisfying the
condition that the Hessian
F
g
ij
(x
i
, y
j
) =
1
2
F
2
y
i
y
j
is not degenerate.
g
2.1.2. Nonlinear dispersion relations
The nonlinear quadratic element (1) results in a nonlinear dispersion relation
between the frequency and the wave vector k
i
of a light ray (see details, for
instance, in Ref. 5),
2
= c
2
[g
b
i
b
j
k
b
i
k
b
j
]
2
1
1
r
q
b
i
1
b
i
2
b
i
2r
y
b
i
1
y
b
i
2r
[g
b
i
b
j
k
b
i
k
b
j
]
2r
, (2)
where, for simplicity, we consider such a relation in a xed point x
k
= x
k
(0)
, when
g
b
i
b
j
(x
k
0
) = g
b
i
b
j
and q
b
i
1
b
i
2
b
i
2r
= q
b
i
1
b
i
2
b
i
2r
(x
k
0
). The coecients q
b
i
1
b
i
2
b
i
2r
should be
f
This formula holds also in GR, when the local co-ordinates on a (pseudo) Riemannian manifold
are chosen such a way that the coecients of metric g
b
i
b
j
(x
i
) are constrained to be a solution of
Einstein equations. For certain local constructions in vicinity of a point x
i
(0)
, we can omit the
explicit dependence on x
i
and consider only formulas derived for y
i
.
g
It is positively denite for models of Finsler geometry; for pseudo-Finsler congurations,
9,26,33,34
this condition is not imposed. Here we also note that physical implications of Finsler type defor-
mations of SR were analyzed in Refs. 1, 35, 36, 32 and 11.
1250072-4
computed from a model of QG, or from a well dened Finsler like modication
of the GR theory. For a locally anisotropic spacetime aether, i.e. in a modied
classical model of gravity with broken local Lorentz invariance, the coecients for
dispersions of type (2) have be measured following some experiments when light
rays propagate according to a Riemannian/Finsler metric.
Dispersion relations should be parametrized and computed dierently for the-
ories with nonlocal interactions and noncommutative variables. Nevertheless, the
form (2) is a very general one which can be obtained in various Finsler like and
extra dimension models even the values of coecients q
b
i
1
b
i
2
b
i
2r
depend on the class
of exact solutions of certain generalized gravitational equations, types of classical
and quantum models, etc.
2.1.3. Finsler metrics do not dene complete geometric models
Any (pseudo) Riemannian geometry on (for our purposes, we consider any necessary
smooth class) manifold M is determined by a metric eld g = g
ij
(x)e
i
e
i
, where
x={x
i
} label the local co-ordinates and the coecients of a symmetric tensor g
ij
(x)
are dened with respect to a general nonholonomic co-frame e
i
= e
i
i
(x)dx
i
.
h
There
is on M a second fundamental geometric object, the LeviCivita connection, =
{
i
} (parametrized locally by coecients

i
jk
for a one-form

i
j
=

i
jk
(x)dx
k
)
which is completely dened by a set {g
ij
} if and only if we impose two basic
conditions: (1) metric compatibility,
k
g
ij
= 0; (2) zero torsion,

T
i
= e
i
=
de
i
+

i
j
e
j
= 0, where is the anti-symmetric product forms (see, for instance,
Ref. 37).
Contrary to (pseudo) Riemannian geometry completely determined by a
quadratic linear form (a metric), a Finsler metric (1) does not state a geometric
spacetime model in a self-consistent and complete form. An element ds = F(x
i
, y
j
)
and Hessian
F
g
ij
(x
i
, y
j
) do not dene completely any metric and connections struc-
tures on the total space TM of a tangent bundle (TM, , M) , where is a surjective
projection (see Refs. 19 and 23).
There are necessary additional suppositions in order to elaborate a well-
dened spacetime model generated by F(x
i
, y
j
), i.e. a Finsler gravity theory. Such
a locally anisotropic gravity is determined by three fundamental geometric objects
on TM and TTM, a metric structure,
F
g, a nonlinear connection,
F
N, and a linear
connection,
F
D, which is adapted to
F
N.
i
h
We follow the system of notations from Refs. 10, 22, 26 and 21 when underlined, primed and
other type indices are used in order to distinguish, for instance, the local co-ordinate co-base dx
i
,
with dx
i
dx
j
dx
j
dx
i
= 0, and an arbitrary one, e
i
, for which certain nonholonomy (equivalently,
anholonomy or nonintegrability) conditions are satised, e
i
e
j
e
j
e
i
= w
ij
k
e
k
, with w
ij
k
being the
anholonomy coecients. For simplicity, we shall omit priming/underling of indices if that will not
result in ambiguities.
i
We put a left label F in order to emphasize (if necessary) that some objects are introduced for a
Finsler geometry model.
1250072-5
S. I. Vacaru
2.2. Finsler geometry on nonholonomic spacetimes
We consider the main concepts and fundamental geometric objects which are nec-
essary for Finsler geometry, and gravity, models on nonholonomic tangent bun-
dles/manifolds (spacetimes).
2.2.1. Fundamental geometric objects in Finsler geometry
Nonlinear connections: A rigorous analysis of the nonlinear connection struc-
tures
F
N was many times omitted by physicists in their works on Finsler gravity
and analyses of experimental restrictions on velocity dependent theories.
1,2
This
geometric/physical object is less familiar to researches working in particle physics
and cosmology and it is confused with nonlinear realizations of connections for
generalized gauge theories.
A nonlinear connection (N-connection) N can be dened as a Whitney sum
(equivalently, a nonholonomic distribution):
TTM = hTM vTM, (3)
with a conventional splitting into horizontal (h), hTM and vertical (v), vTM,
subspaces. It is given locally by a set of coecients N = {N
a
i
}, when
j
N =
N
a
i
(u)dx
i

y
a
. There is a frame (vielbein) structure which is linear on N-
connection coecients and on partial derivatives
i
= /x
i
and
a
= /y
a
and, respectively, their duals, dx
i
and dy
a
,
e
= (e
i
=
i
N
a
i

a
, e
a
=
a
), (4)
e
= (e
i
= dx
i
, e
a
= dy
a
+N
a
i
dx
i
). (5)
The vielbeins (5) satisfy the nonholonomy relations [e
, e
] = e
=
w
with (antisymmetric) nontrivial anholonomy coecients w

b
ia
=
a
N
b
i
and
w
a
ji
=
a
ij
, where
a
ij
= e
j
(N
a
i
) e
i
(N
a
j
) are the coecients of N-connection
curvature.
k
For a TM endowed with a generating Finsler function F, we can introduce a
homogeneous Lagrangian L = F
2
. There is the canonical (Cartans) N-connection
with coecients:
c
N
a
i
=
G
a
y
n+i
, for G
a
=
1
4
F
g
a n+i

2
L
y
n+i
x
k
y
n+k
L
x
i
, (6)
where
F
g
ab
is inverse to
F
g
ab
.
l
j
Co-ordinates u = (x, y) on an open region U TM are labeled in the form u
= (x
i
, y
a
), with
indices of type i, j, k, . . . = 1, 2, . . . , n and a, b, c = n+1, n+2, . . . , n+n; on TM, x
i
and y
a
are
respectively the base co-ordinates and ber (velocity like) co-ordinates; we use boldface symbols
for spaces (and geometric objects on such spaces) enabled with N-connection structure.
k
The holonomic/integrable frames are selected by the integrability conditions w
= 0.
l
Respective contractions of h- and v-indices, i, j, . . . and a, b, . . . , are performed following the rule:
we write an up v-index a as a = n + i and contract it with a low index i = 1, 2, . . . , n; on
1250072-6
For any set N
a
i
, we can chose a well-dened F and corresponding frame coef-
cients e
and (inverse) e

when e
transform N
a
i
into respective
c
N
a
i
(for instance, N
a
i
= e
i
i
e
a
a
c
N
a
i
, or using more general types of transforms).
So, we can work equivalently with any convenient set N = {N
a
i
}, which may be
redened as a canonical set
F
N
a
i
=
c
N
a
i
.
In our works devoted to applications of Finsler geometry methods in modern
gravity and string theory
10,22,26
related to standard physics models, we considered
N-connections not only on tangent bundles but also on nonholonomic manifolds (see
denition in footnote b). In such approaches, it is considered a general manifold V,
instead of TM, when the Whitney sum (3), existing naturally on vector/tangent
bundles, is introduced as a nonholonomic distribution on V with conventional
h- and v-splitting into (holonomic and nonholonomic variables, respectively, dis-
tinguished by co-ordinates x
i
and y
a
),
TV = hVvV. (7)
A N-anholonomic manifold (or tangent bundle; in brief, we shall write respec-
tively the terms bundle and manifold; we can consider similarly vector bundles)
is a nonholonomic manifold enabled with N-connection structure (7). The prop-
erties of a N-anholonomic bundle/manifold are determined by N-adapted bases
(4) and (5). A geometric object is N-adapted (equivalently, distinguished), i.e.
it is a d-object, if it can be dened by components adapted to the splitting (7)
(one uses terms d-vector, d-form, d-tensor). For instance, a d-vector is represented
as X = X
= X
i
e
i
+ X
a
e
a
and a one d-form

X (dual to X) is represented as
X = X
= X
i
e
i
+X
a
e
a
.
m
total spaces of even dimensions, we can write y
i
instead of y
n+i
or y
a
. The spacetime signature
may be encoded formally into certain systems of frame (vielbein) coecients and co-ordinates,
some of them being proportional to the imaginary unity i, when i
2
= 1. For a local tangent
Minkowski space of signature (, +, +, +), we can chose e
0
= i/u
0
, where i is the imaginary

unity, i
2
= 1, and write e
= (i/u
0
, /u
1
, /u
2
, /u
3
). Euclidean co-ordinates with

i were used in textbooks on relativity theory (see, for instance, Refs. 38 and 39). Later, they
were considered for analogous modeling of gravity theories as eective Lagrange mechanics, or
Finsler like, geometries.
10,26
The term pseudo-Finsler was also introduced in a dierent form,
more recently, for some analogous gravity models (see, for instance, Ref. 9) and a mathematical
book.
40
m
A geometric object can be redened equivalently for arbitrary frame and co-ordinate systems;
nevertheless, the N-adapted constructions allow us to preserve a prescribed h- and v-splitting. In
this paper, we omit details on co-ordinate transforms of geometric objects (for instance, for the
coecients of linear and linear connections), on vector/tangent bundles or nonholonomic mani-
folds, which are considered in Refs. 19, 10, 22 and 26. Geometrically, all formulas are similar and
do not depend on the fact what type of Whitney sum, we use (7) or (3). There are dierences
depending on the type of physical theory we model. For instance, on TM, the v-components
can be related to some velocity components, but on V such v-components are distinguished
by nonintegrable constraints encoded into the nonholonomic frame structure. The hv-splitting
exists naturally (it can be a holonomic or anholonomic one, depending on the type of geomet-
ric/physical model we consider) on any vector bundle. Such a splitting can be modeled as a local
bered structure on any (for instance, pseudo-Riemannian) manifold by introducing correspond-
ing classes of nonholonomic frames. We can introduce the h- and v-decompositions in the same
1250072-7
S. I. Vacaru
Lifts of base metrics on total spaces: The most known procedure to extend
F
g
ab
to a metric in TM is the so-called Sasaki type lift when the Hessian metric is
considered in N-adapted form both of the h- and v-components metric,
F
g =
F
g
du
du
=
F
g
ij
(u)dx
i
dx
j
+
F
g
ab
(u)
c
e
a
c
e
b
,
c
e
a
= dy
a
+
c
N
a
i
(u)dx
i
, du
= (dx
i
, dy
a
).
(8)
Similarly, we can dene a metric structure for an even dimensional N-anholonomic
manifold V (this condition is satised for any TM) endowed with a h-metric g
ij
,
on hV, and a given set of N-connection coecients N
a
i
.
Using arbitrary frame transforms with coecients e
(u), we can transform

the total Finsler metric (8) into a general one g = g
on TM, where
g
= e
F
g
and e
= e
(u)
c
e
, for
c
e
= (dx
i
,
c
e
a
). A metric d-tensor
(d-metric) is with n + n splitting. Such a N-adapted decomposition can be per-
formed by corresponding parametrizations of components of matrices e
, when
g
= [g
ij
, h
ab
, N
a
i
]. Having redened the co-ordinates and frame coecients, we
can express:
g = g
ij
(x, y)dx
i
dx
j
+h
ab
(x, y)e
a
e
b
,
e
a
= dy
a
+N
a
i
(x, y)dx
i
.
(9)
With respect to local dual co-ordinate frames, metric (9) is parametrized:
g = g
(u)du
du
, (10)
g
g
ij
+N
a
i
N
b
j
h
ab
N
e
j
h
ae
N
e
i
h
be
h
ab
. (11)
We emphasize that the values N
a
i
(u) should not be identied as certain gauge
elds in a KaluzaKlein theory if we do not consider compactications on
co-ordinates y
a
.
n
For simplicity, hereafter we shall work on a general nonholonomic space V
enabled with N-connection splitting N (7) and resulting N-adapted base and
co-base (4) and (5). Such a space is also endowed with a symmetric metric struc-
ture g (of necessary local Euclidean or pseudo-Euclidean signature) which can be
parametrized in the form (9) (or, equivalently, (10)). Any metric g on V can be rep-
resented equivalently in the form
F
g (8) after corresponding frame and co-ordinate
transforms for a well dened generating function F(x, y). It is always possible to
introduce on a (pseudo) Riemannian manifold/tangent bundle V some local Finsler
way on TV and/or TTM xing a corresponding N-connection structure, with very similar rules
of transforms of geometric/physical objects. Nevertheless, physical meaning of such objects are
completely dierent for constructions on tangent bundles and nonholonomic manifolds.
n
In Finsler like theories, a set {N
a
i
} denes a N-connection structure, with elongated partial
derivatives (4). In KaluzaKlein gravity they are linearized on y
a
, N
a
i
= A
a
bi
(x)y
b
, when A
a
bi
(x)
are treated as some YangMills potentials inducing covariant derivatives.
1250072-8
like variables when the metric is parametrized in a Sasaki type form. We shall write
V = TM when it will be necessary to emphasize that the constructions are dened
explicitly for tangent bundles.
Distinguished connections, theirs torsions and curvatures: For any d-
metric of type (8) and/or (9), we can construct the LeviCivita connection =
{
} on V in a standard form. Nevertheless, this connection is not used in Finsler

geometry and generalizations. The problem is that is not compatible with a N-
connection splitting, i.e. under parallel transports with , it does not preserve the
Whitney sum (7).
In order to perform geometric constructions with h-/v-splitting, it was
introduced the concept of distinguished connection (in brief, d-connection). By
denition, such a d-connection D = (hD, vD) is a linear one preserving under par-
allelism the N-connection structure on V. The N-adapted components
of a
d-connection D are dened by equations D
and parametrized in the

form
= (L
i
jk
, L
a
bk
, C
i
jc
, C
a
bc
), where D
= (D
i
, D
a
), with hD = (L
i
jk
, L
a
bk
) and
vD = (C
i
jc
, C
a
bc
) dening the covariant, respectively, h- and v-derivatives.
The simplest way to perform computations with a d-connection D is to associate
it with a N-adapted dierential one-form
, when the coecients of

forms and tensors (i.e. d-tensors etc.) are dened with respect to (5) and (4). In
this case, we can apply the well known formalism of dierential forms as in GR.
37
It also allows us to elaborate an N-adapted dierential/integral calculus for Finsler
spaces and generalizations. For instance, torsion of D is dened/computed
T
De
= de
, (12)
see formulas (A.1) in Appendix, for explicit values of coecients T
= {T
}.
Similarly, using the d-connection one-form (4), one computes the curvature of D
(d-curvature)
R
= d
= R
, (13)
see formulas (A.2) for h-v-adapted components, R
= {R
}.
The Ricci d-tensor Ric = {R
} is dened in a standard form by contracting

respectively the components of (A.2), R
, The h-/v-components of this

d-tensor, R
= {R
ij
, R
ia
, R
ai
, R
ab
}, are,
R
ij
R
k
ijk
, R
ia
R
k
ika
, R
ai
R
b
aib
, R
ab
R
c
abc
, (14)
see explicit coecients formulas (A.2). Here we emphasize that for an arbitrary
d-connection D, this tensor is not symmetric, i.e. R
= R
. In order to dene
the scalar curvature of a d-connection D, we have to use a d-metric structure g (9)
on V, or TM,
s
R g
= g
ij
R
ij
+h
ab
R
ab
, with R = g
ij
R
ij
and S = h
ab
R
ab
being respectively the h- and v-components of scalar curvature.
1250072-9
S. I. Vacaru
For any d-connection D in Finsler geometry, and generalizations, the Einstein
d-tensor is (by denition):
E
1
2
g
s
R. (15)
This d-tensor is also not symmetric and, in general, D
= 0. Such a tensor
is very dierent from that for the LeviCivita connection which is symmetric
and with zero covariant divergence, i.e. E
= E
and
= 0, where E
is
computed using

.
2.2.2. Notable connections for Finsler spaces
In general, it is possible to dene on V two independent fundamental geometric
structures g and D which are adapted to a given N. For applications in modern
physics, it is more convenient to work with a d-connection D, which is metric
compatible satisfying the condition Dg = 0 (see discussions in Refs. 10, 22 and 26).
There is an innite number of d-connections which are compatible to a metric g.
A special interest presents a subclass of such metrics which are completely dened
by g in a unique N-adapted form following a well dened geometric principle.
The canonical d-connection: In our works on Finsler gravity, we used the so-
called canonical d-connection

D (on spaces of even dimensions it is called the
h-/v-connection, such connections were studied geometrically in Ref. 19 on vec-
tor/tangent bundles and for generalized Finsler geometry).
By denition,

D is with vanishing horizontal and vertical components of tor-
sion and satises the conditions

Dg = 0 (see explicit component formulas (A.3)
and (A.1)). From many points of view, on a nonholonomic space V,

D is the
best N-adapted analog of the LeviCivita connection . We have the distortion
relation:
=

D+

Z, (16)
when both linear connections = {
} and

D = {
} and the distorting

tensor

Z = {
} are uniquely dened by the same metric tensor g. The coef-

cient formulas are given in Appendix (see (A.4) and (A.5)). The connection

D
is with nontrivial torsion (the coecients

T
i
ja
,

T
a
ji
and

T
a
bi
are not, in general,
zero; see (A.1)) but such d-torsions are nonholonomically induced by N-connection
coecients and completely determined by certain o-diagonal N-terms in (11). All
geometric constructions can be performed equivalently and redened in terms of
both connections and

D using (16).
The connection

Dand various types of d-connections D = (hD, vD) with h- and
v-covariant derivatives, hD = (L
i
jk
, L
a
bk
) and vD = (C
i
jc
, C
a
bc
) can be dened on vec-
tor bundles and on (pseudo) Riemann spaces of arbitrary dimensions, alternatively
to . A general Finsler d-connection is of type
F
D = {L
i
jk
, C
i
jc
} with coecients
1250072-10
L
i
jk
and C
i
jc
determined by a generating Finsler function F and an N-connection
N = {N
a
i
} and some arbitrary and/or induced nonholonomically torsion elds. In
general,
F
D is not metric compatible, i.e.
F
D
F
g =
F
Q = 0.
Nonmetric Finsler d-connections: There were considered three types of such
notable connections (see details in Refs. 19 and 23 and references therein and,
on physical applications, Refs. 10 and 22). In general, it is possible to dene fol-
lowing dierent geometric principles an innite number of metric noncompatible
or compatible d-connections in Finsler geometry and generalizations. The rst
(metric noncompatible) one was the Berwald d-connection
F
D =
B
D = {L
a
bk
=
N
a
k
/y
b
, C
i
jc
= 0}. It is completely dened by the N-connection structure and
B
D
F
g =
B
Q = 0.
Then, the Chern d-connection was introduced,
23 F
D =
Ch
D =

L
i
jk
, C
i
jc
= 0,
where

L
i
jk
is given by the rst formula in (A.3). This d-connection is torsionless,
Ch
T = 0, but (in general) metric noncompatible,
Ch
D
F
g=
Ch
Q = 0. Recently,
some authors attempted to elaborate ChernFinsler, or BerwaldFinsler cosmolog-
ical models and certain modications of Einstein gravity using such metric non-
compatible d-connections.
13,14
Geometrically, the Chern d-connection is a nice
one but with a number of problems for applications in standard models of physics
(because of nonmetricity, there are diculties in denition of spinors and Dirac
operators, conservation laws, quantization, etc.; see critical remarks in Refs. 6
and 10).
o
A preferred metric compatible Cartan d-connection: Historically, it was the
rst d-connection introduced in Finsler gravity
16
in 1935. Perhaps, the rst model of
Finsler gravity with the Einstein equations formulated for the Cartan d-connection
was proposed by Horvath
17
in 1950. Later, such constructions were generalized
for other classes of d-connections on vector/tangent bundles and nonholonomic
manifolds (see details in Refs. 19, 10 and 22). It is given by local coecients
F
D =
c
D = {
c
L
i
ik
,
c
C
a
bc
}, where,
c
L
i
jk
=
1
2
F
g
ir
(e
k
F
g
jr
+e
j
F
g
kr
e
r
F
g
jk
),
c
C
a
bc
=
1
2
F
g
ad
(e
c
F
g
bd
+e
c
F
g
cd
e
d
F
g
bc
).
(17)
The Cartan d-connection (17) is metric compatible,
c
D
F
g = 0, but with nontrivial
torsion
c
T = 0 (the second property follows from formulas (12) and (A.1) redened
o
Experts on Finsler geometry also know about the Hashiguchi d-connection
F
D =
H
D = {L
a
bk
=
N
a
k
/y
b
,
H
C
i
jc
}, where
H
C
i
jc
=
1
2
F
g
ad
(e
c
F
g
bd
+e
c
F
g
cd
e
d
F
g
bc
), for e
c
= /y
c
and a given
F
g
bd
(see details, for instance, in Ref. 19). It contains both nontrivial torsion and nonmetricity
components, all completely dened by the N-connection and Hessian structure. We studied Finsler-
ane theories with very general torsion and nonmetric structures, provided examples of exact
solutions and discussed physical implications of models in Part I of the monograph in Ref. 22.
1250072-11
S. I. Vacaru
for
c
D). The nontrivial torsion terms are induced nonholonomically by a fundamen-
tal Finsler function F via
F
g
bd
and Cartans N-connection
c
N
a
i
(6). This torsion
is very dierent from the well known torsion in EinsteinCartan, or string/gauge
gravity models, because in the CartanFinsler case we do not need additional eld
equations for the torsion elds. The torsion
c
T , similarly to

T , is completely dened
by a (Finsler) d-metric structure.
p
In Finslerian theories of gravity, it is possible to work geometrically with very
dierent types of d-connections because there are always some transformations of
the Cartan d-connection into any mentioned above (or more general ones) notable
Finsler connections, of Berwald, Chern or Hashiguchi types. Nevertheless, the main
issue is that for what kind of d-connections we can formulate Einstein/Dirac/
YangMills etc equations which are well dened, self-consistent and with important
physical implications. In our approach, for gravity models on nonholonomic mani-
folds/bundles of arbitrary dimension, we give priority to the canonical d-connection
D (here we also note that the Einstein equations for this d-connection can be inte-
grated in very general forms
21,22,26,33,34
).
3. Field Equations for Finsler Theories of Gravity
In this section, we outline the theory of EinsteinFinsler spaces. Einstein equations
are formulated for the canonical and/ or Cartan d-connections. A class of theo-
ries extending the four dimensional GR to metric compatible Finsler gravities on
tangent bundles and/or nonholonomic manifolds are analyzed. We end with a dis-
cussion of principles of GR and their extension to metric compatible Finsler gravity
theories.
3.1. Einstein equations for distinguished connections
3.1.1. Gravitational eld equations in h-/v-components
Having prescribed a N-connection N and d-metric g (8) structures on a
N-anholonomic manifold V, for any metric compatible d-connection D, we can
compute the Ricci R
(14) and Einstein E
(15) d-tensors. The N-adapted grav-

itational eld equations
10,19,21,22
are E
, where the source
has to be
dened in explicit form following certain explicit models of locally anisotropic
gravitational and matter eld interactions.
q
p
A very important property of
c
D is that it denes also a canonical almost symplectic connec-
tion
18,19
(see details and recent applications to quantization of Finsler spaces and Einstein/brane
gravity in Ref. 15). Perhaps, the Cartan d-connection is the best one for physical applications in
modern physics of Finsler geometry and related anholonomic deformation method (see additional
arguments in Refs. 6, 10, 22 and 15). The d-connections
c
D and
b
D allow us to work in N-adapted
form in Finsler classical and QG theories keeping all geometric and physical constructions to be
very similar to those for the LeviCivita connection .
q
For theories with arbitrary torsions T , we have to complete such equations with additional
algebraic or dynamical ones (for torsions coecients) like in the EinsteinCartan, gauge or string
1250072-12
The eld equations for metric compatible d-connections in Finsler gravity the-
ories can be distinguished in the following form:
R
ij

1
2
(R +S)g
ij
=
ij
, (18)
R
ab
1
2
(R +S)h
ab
=
ab
, (19)
R
ai
=
ai
, R
ia
=
ia
, (20)
where R
ai
= R
b
aib
and R
ia
= R
k
ikb
are dened by formulas for d-curvatures (A.2)
containing d-torsions (A.1). For a metric compatible d-connection D, which is
completely dened by a d-metric structure g, the corresponding system (18)(20)
is very similar to that for the usual Einstein gravity. The dierence is that R
ai
=
R
ia
, = D.
r
3.1.2. Equations for

D and
c
D
Because the canonical d-connection

D is completely dened by g
, the correspond-
ing Finsler analog of Einstein, we use the tensor (15) for D =

D,
1
2
g
s
R =

, (21)
can be constructed to be equivalent to the Einstein equations for . This is possible
if

=
matter
+
z
are derived in such a way that they contain contributions

from (1) the N-adapted energymomentum tensor (dened variationally following
the same principles as in GR but on V) and (2) the distortion of the Einstein tensor
in terms of

Z (16), i.e. (A.4),

E
=

E
+
z
, for
z
=
z
. The value
z
is computed by introducing

D =
Z into (A.2).
The system of Eqs. (A.2) can be integrated in very general forms, see explicit
constructions in Sec. 3.1.4. Such solutions can be considered also in GR if we impose
additionally the condition that
L
c
aj
= e
a
(N
c
j
),

C
i
jb
= 0,
a
ji
= 0, (22)
theories with torsion. On generalized Finsler spaces, such constructions should be in N-adapted
forms (see details in Part I of Ref. 22). In brief, we note here that the N-adapted tensor, covariant
dierential/integral calculus can be performed very similarly to the well known tetradic formalism,
in our case, using respectively, the N-elongated partial derivatives and dierentials, (4) and (5).
This way, we can elaborate a N-adapted variational calculus on TM, or V using the corresponding
d-connection and d-metric structures. For metric compatible d-connections, in N-adapted bases,
all constructions are very similar to those in GR. We can provide proofs for locally anisotropic
uid/spinning models, Dirac equations, YangMills elds on Finsler spaces, etc. (see details in
Refs. 10 and 22). This is the priority of the canonical/Cartan d-connection structure. We cannot
generate simple physical theories if we work with the metric noncompatible Chern d-connection
(see additional critics and discussion in Refs. 6 and 41).
r
Elaborating geometric/gravity models on TM, containing in the limit D the Einstein gravity
theory on M, we should consider that Eq. (18) dene a generalization of

R
ij
1
2

Rg
ij
=

ij
for
= {
i
jk
} and a well-dened procedure of compactication, or brane like warping/trapping
on y
a
. The observed three-dimensional space, with possible Finsler type contributions, is contained
in such classes of solutions.
1250072-13
S. I. Vacaru
for
(matter energymomentum in Einstein gravity) if

D . We
emphasize here that if the constraints (22) are satised the tensors

T
(A.1) and
Z
(A.5) are zero. For such congurations, we have

, with respect to
(4) and (5), see (A.4), even

D = .
There are two very important benets to work with canonical Finsler vari-
ables (for instance with the Cartan d-connection) on (pseudo) Riemannian man-
ifolds: (1) the Einstein equations magically separate and can be integrated in
very general forms
21,26,33
; (2) there is also an equivalent almost K ahler representa-
tion for such Finsler variables in GR, which allows us to perform various types of
deformation/A-brane quantization and/or two connection renormalization.
6,15,42,43
In Finsler geometry/gravity models, the constraints (22) are not obligatory. On
TM, and any even dimensional V, it is possible to perform such frame deforma-
tions when

D
c
D. So, the Einstein equations for the Cartan d-connection, in
GR and Finsler generalizations, also can be integrated in very general forms.
An extra dimensional gravity theory can be elaborated for a linear connec-
tion (in general, it can be metric noncompatible) with y
a
are considered as extra
dimension co-ordinates to a 4-d (pseudo) Riemannian spacetime manifold with
co-ordinates x
i
. Standard Finsler theories are elaborated on tangent bundles with
nontrivial N-connection structure (when y
a
are typical ber co-ordinates which
can be identied with certain velocity elds if sections on basic manifolds are
considered). In general, Finsler like variables can be introduced on arbitrary man-
ifolds, or bundle spaces (in particular, parametrizing exact solutions in Einstein
gravity; see detailed discussions and examples in Refs. 10, 15, 33 and 26). In GR,
the (Finsler like) N-connection coecients parametrize certain classes of nonholo-
nomic frames and o-diagonal metrics when co-ordinates y
a
are considered for a
2+2 splitting (but not as some velocity type variables). The concept of nonholo-
nomic manifold allows us to formulate an unied geometric approach for all classes
of such geometric and physical models.
3.1.3. EinsteinFinsler spaces
An Einstein manifold (space) is dened in standard form by a LeviCivita con-
nection = {
} satisfying the eld equations

R
ij
= g
ij
, for a (pseudo)
Riemannian manifold M endowed with metric g
ij
, where is the cosmological
constant. In order to apply the anholonomic frame method for constructing exact
solutions from higher dimension gravity,
33
we should introduce three shells of
anisotropy.
s
For a d-connection D which is metric compatible with g (9) on V,
s
Parametrization I with anisotropic shells for higher-order anisotropic extensions, see
details in Ref. 33, when TV=hTV
0
vTV
1
vTV
2
vTV, for local co-
ordinates u
1
=(u
0
, u
1
a
) =(u
i
, u
0
a
, u
1
a
), u
2
=(u
1
, u
2
a
) =(u
i
, u
0
a
, u
1
a
, u
2
a
), with
i, j, . . . =1, 2, . . . , n;
0
a,
0
b, . . . =n + 1, . . . , n + m;
1
a,
1
b, . . . =n + m + 1, . . . , n + m + +
1
m;
2
a,
2
b, . . . =n + m + 1, . . . , n + m + +
1
m, n + m + +
1
m + +
2
m. We can
1250072-14
we can consider generalized Einstein spaces dened by
R
ij
=
h
(u)g
ij
, R 0
a
0
b
=
0
v
(u)h 0
a
0
b
,
R 1
a
1
b
=
1
v
(u)h 1
a
1
b
, R 2
a
2
b
=
2
v
(u)h 2
a
2
b
,
R
a
0
i
= R
i
0
a
= 0, R 2
ai
= R
i
2
a
= 0, R 1
ai
= R
i
1
a
= 0,
R 2
a
1
a
= R 1
a
2
a
= 0, R 2
a
0
a
= R 0
a
2
a
= 0,
(23)
where
h
(u) and
0
v(u),
1
v
(u),
2
v
(u) are respectively the so-called locally
anisotropic h- and
0
v-,
1
v-,
2
v-polarized gravitational constants. Such polariza-
tions should be dened for certain well-dened constraints on matter and gravita-
tional eld dynamics, lifts on tangent bundles, corrections from QG or any extra
dimension gravitational theory.
In this work, an EinsteinFinsler space is dened by a triple [N, g, D] with a
metric compatible d-connection D subjected to the condition to be a solution of
Eq. (23) with sources of type
3
= diag[3
]. Such equations can be solved

in very general form for the canonical d-connection

D and certain nonholonomic
restrictions to the LeviCivita connection .
20,22,33
In a general context, we can
consider that an EinsteinFinsler space is determined by a set of solutions of (21)
with given sources and for a triple [N, g, D] when D =

D, or
c
D.
3.1.4. Metric ansatz and partial dierential equations
Third-order anisotropic ansatz: Any metric (9) can be reparametrized in a
form with three shell anisotropy,
g = g
ij
(x)dx
i
dx
j
+h 0
a
0
b
(x,
0
y)e
0
a
e
0
b
+h 1
a
1
1
b
(x,
0
y,
1
y)e
1
a
e
1
b
+h 2
a
2
b
(x,
0
y,
1
y,
2
y)e
2
a
e
2
b
,
e
0
a
= dy
0
a
+N
0
a
i
(
0
u)dx
i
, e
1
a
= dy
1
a
+N
1
a
i
(
1
u)dx
i
+N
1
a
0
a
(
1
u) e
0
a
,
e
2
a
= dy
2
a
+N
2
a
i
(
2
u)dx
i
+N
0
a
0
a
(
2
u) e
0
a
+N
2
a
1
a
(
2
u) e
1
a
,
(24)
for x = {x
i
},
0
y = {y
0
a
},
1
y = {y
1
a
},
2
y = {y
2
a
}, when the vertical indices
and co-ordinates split in the form y = [
0
y,
1
y,
2
y], or y
a
= [y
0
a
, y
1
a
, y
2
a
];
0
u = (x,
0
y),
1
u = (
0
u,
1
y),
2
u = (
1
u,
2
y), or u
0
= (x
i
, y
0
a
), u
1
=
(u
0
, y
1
a
), u
2
= (u
1
, y
2
a
).
consider n=4 and m =
1
m =
2
m = 2, when dimM = 2 or 4, dimV = 8. Parametrization II is
for n = 4, m = 4,
1
m = 0, with trivial shall (u
1
a
) and local co-ordinates u
= (x
i
, y
a
), for
i = 1, 2, 3, 4 and consequently a = 5, 6, 7, 8 when on tangent bundles 5 can be contracted to 1, 6
to 2 and so on. Parametrization III is for n = 2 and m = 2, when dimM = 2, dimV = 4, with
indices i = 1, 2 and a = 3, 4.
1250072-15
S. I. Vacaru
There is a very general ansatz of this form (with Killing symmetry on y
8
, when
the metric coecients do not depend on variable y
8
; it is convenient to write y
3
=
0
v,
y
5
=
1
v, y
7
=
2
v and introduce parametrization of N-coecients via n- and w-
functions) dening exact solutions of (23),
sol
g = g
i
(x
k
)dx
i
dx
i
+h 0
a
(x
k
,
0
v)e
0
a
e
0
a
+h 1
a
(u
0
,
1
v) e
1
a
e
1
a
+h 2
a
(u
1
,
2
v) e
2
a
e
2
a
,
e
3
= dy
3
+w
i
(x
k
,
0
v)dx
i
, e
4
= dy
4
+n
i
(x
k
,
0
v)dx
i
,
e
5
= dy
5
+w 0
(u
0
,
1
v)du
, e
6
= dy
6
+n 0
(u
0
,
1
v)du
0
,
e
7
= dy
7
+w 1
(u
1
,
2
v)du
1
, e
8
= dy
8
+n 1
(u
1
,
2
v)du
1
.
(25)
In Theorem 1.1 of Ref. 33 (we should consider those results for three shells and
trivial -coecients), there are stated explicit conditions on w- and n-coecients
and -sources, for arbitrary dimensions and in very general forms, when an ansatz
(25) generates exact solutions gravity.
Separation of equations for the canonical d-connection: Let us consider
the ansatz (25) for dimensions n = 2 and
0
m = 2,
1
m =
2
m = 0, when the
source is parametrized in the form
= diag[
;
1
=
2
=
2
(x
k
);
3
=
4
=
4
(x
k
, y
3
)]. Computing the corresponding coecients of d-connection

D following
formulas (A.3) and introducing them respectively into (A.2) (14), we express the
gravitational eld Eq. (21) in the form
t
:
R
1
1
=

R
2
2
=
1
2g
1
g
2
2

g
1
g
2
2g
1
(g
2
)
2
2g
2
+g
1

g
1
g
2
2g
2
(g
1
)
2
2g
1
=
2
(x
k
), (26)
R
3
3
=

R
4
4
=
1
2h
3
h
4
4

(h
4
)
2
2h
4
3
h
4
2h
3
=
4
(x
k
, y
3
), (27)
R
3k
=
w
k
2h
4
4

(h
4
)
2
2h
4
3
h
4
2h
3
+
h
4
4h
4
k
h
3
h
3
+

k
h
4
h
4

k
h
4
2h
4
= 0, (28)
R
4k
=
h
4
2h
3
n
k
+
h
4
h
3
h
3
2
h
k
2h
3
= 0. (29)
In the above formulas, we denote a
= a/x
1
, a
= a/x
2
, a
= a/y
3
.
The system (26)(29) is nonlinear and with partial derivatives. Nevertheless, the
existing separation of equations (we should not confuse with separation of variables
t
The details of such computations can be found in Part II of Ref. 22 and in Ref. 33.
1250072-16
which is a dierent property) allows us to construct very general classes of exact
solutions (depending on conditions if certain partial derivatives are zero, or not). For
any prescribed
2
(x
k
), we can dene g
1
(x
k
) (or, inversely, g
2
(x
k
)) for a given g
2
(x
k
)
(or, inversely, g
1
(x
k
)) as an explicit, or nonexplicit, solution of (26) by integrating
two times on h-variables. Similarly, taking any
4
(x
k
, y
3
), we solve (27) by inte-
grating one time on y
3
and dening h
3
(x
k
, y
3
) for a given h
4
(x
k
, y
3
) (or, inversely,
by integrating two times on y
3
and dening h
4
(x
k
, y
3
) for a given h
3
(x
k
, y
3
)).
Haven determined the values g
i
(x
k
) and h 0
a
(x
k
, y
3
), we can compute the coef-
cients of N-connection: The functions w
j
(x
k
, y
3
) are solutions of algebraic equa-
tion (28). Finally, we have to integrate two times on y
3
in order to obtain n
j
(x
k
, y
3
).
Such general solutions depend on integration functions depending on co-ordinates
x
k
. In physical constructions, we have to consider well-dened boundary conditions
for such integration functions.
Equations for the h-v/Cartan d-connection: In N-adapted frames, the h-v
d-connection

D is determined by coecients
= (
L
a
bk
,

C
a
bc
),

L
i
jk
=
1
2
g
ih
(e
k
g
jh
+e
j
g
kh
e
h
g
jk
),
C
a
bc
=
1
2
h
ae
(e
b
h
ec
+e
c
h
eb
e
e
h
bc
),
(30)
which are computed for a d-metric g = [g
ij
, h
ab
] (9). Via frame transforms to
F
g (8),
g
= e
F
g
, we can dene such N-adapted frames when the coecients of
D are equal to the coecients of the Cartan d-connection

c
D = {
c
L
i
ik
,
c
C
a
bc
} (17).
For dimensions n = 2 and
0
m = 2,
1
m =
2
m = 0, the Eq. (23) for

(30) and
source transform into an exactly integrable system of partial dierential equations
when the coecients of (9) are stated by an ansatz g
= diag[g
i
(x
k
), h
a
(x
i
, v)] and
N
3
k
= w
k
(x
i
, v), N
4
k
= n
k
(x
i
, v). The rst two equations are equivalent, respectively,
to (26) and (27),
R
1
1
=

R
2
2
=

R
1
1
=

R
2
2
=
2
(x
k
), (31)
R
3
3
=

R
4
4
=

R
3
3
=

R
4
4
=
4
(x
k
, y
3
). (32)
Instead of (28), (29) we get, correspondingly,
R
3j
=
h
3
2h
3
w
j
+A
w
j
+B
j
= 0, (33)
R
4i
=
h
4
2h
3
n
i
+
h
4
2
K
i
= 0, (34)
where
A =
3
2h
3
+
h
4
2h
4
, B
k
=
h
4
2h
4
k
g
1
2g
1

k
g
2
2g
2
k
A,
K
1
=
1
2
1
g
2
h
3
+
g
2
g
2
h
4
, K
2
=
1
2
2
g
1
h
3
2
g
2
h
4
.
(35)
1250072-17
S. I. Vacaru
The system (31)(34) also has the property of separation of equations. In this
case (having dened h
a
(x
i
, v)), we can compute n
k
(x
i
, v) integrating Eq. (34) on
y
3
=v and, respectively, solving an usual rst-order dierential equation (33), on
y
3
=v, considering x
i
as parameters. Prescribing a generating function F(x
i
, v),
such a solution given by data g
i
(x
k
), h
a
(x
i
, v) and N
3
k
=w
k
(x
i
, v), N
4
k
=n
k
(x
i
, v)
can be represented equivalently as a (pseudo) Finsler space. To associate such a
h-v-conguration to a real Finsler geometry is convenient to work with sets of local
carts on TM, or V, when the quadratic algebraic system for e
has well dened

real solutions.
General solutions for Finsler gravity: The system of gravitational eld equa-
tions with three shell anisotropy (when its rst shell restriction given by (26)(29))
can be solved in general form following the results of the above mentioned Theo-
rem 1.1 from Ref. 33. We, in this work, omit such cumbersome formulas (3d-order
solutions for anisotropic canonical d-connections are considered in Appendix B) but
give an explicit example of anisotropic generalization of FriedmannRobertson
Walker (FRW) solutions depending on time and three velocity co-ordinates in
Sec. 4.2.
In this subsection, we provide the general solution of Eqs. (31)(34) for the
h-v/Cartan d-connection. It can be written for a d-metric g = [g
ij
, h
ab
] (9) with
coecients computed in the following form:
g
i
=
i
e
(x
k
)
, for
1
+
2
=
2
(x
k
),
h
3
=
3
0
h(x
i
) [f
(x
i
, v)]
2
|(x
i
, v)|,
for =
0
(x
i
)

3
8
0
h(x
i
)
dv
4
(x
k
, v)f
(x
i
, v) [f(x
i
, v)
0
f(x
i
)],
h
4
=
4
[f(x
i
, v)
0
f(x
i
)]
2
,
w
j
=
0
w
j
(x
i
) exp
v
0
2h
3
A
vv
1
dv
1
v
0
dv
1
h
3
B
j
h
vv
1
exp
v
1
0
2h
3
A
vv
1
dv
1
, n
i
=
0
n
i
(x
k
) +
dv h
3
K
i
.
(36)
Such solutions with h
3
= 0 and h
4
= 0 are determined by generating functions
f(x
i
, v), f
= 0, and integration functions

0
f(x
i
),
0
h(x
i
),
0
w
j
(x
i
) and
0
n
i
(x
k
); the
coecients A and B
j
, K
i
are given by formulas (35).
u
u
We can prove that such coecients generate general solutions by straightforward computations
being similar to those for the canonical d-connections (see details in Refs. 22 and 33).
1250072-18
3.2. Principles of GR and Finsler gravity
The goal of this section is to show how the theoretic scheme and principles for GR
can be extended to metric compatible Finsler gravity theories.
At present, all classical gravitational phenomena completely described the stan-
dard GR (one may be some exceptions for dark energy and dark matter theories; we
shall discuss the problem in Sec. 4). However, it is generally accepted that the incom-
patibility between GR and quantum theory should be treated in a more complete
theory of QG which is under elaboration. It is supposed also that from any general
theory (string/brane gravity, commutative and noncommutative gauge or other gen-
eralizations, various quantum models etc) GR is reproduced with small corrections
in the classical limit. Such small corrections necessarily violate the equivalence
principle in GR.
44
This also results in violation of local Lorenz invariance both
in special relativity (SR) and GR (we explained in Sec. 2.1 how nonlinear dis-
persion relations from any QG model result in Finsler type metrics depending on
velocitiy/momentum variables).
3.2.1. Minimal extensions of Einstein gravity to Finsler theories
Dierent researches in geometry and physics work with dierent concepts of Finsler
space. In order to avoid ambiguities, let us state what rules we follow in this paper
in order to elaborate a class of Finsler gravity theories which seem to be most closed
to the modern paradigm of standard physics.
A (pseudo) Finsler geometry model is canonically dened by a (fundamental
Finsler function F(x, y) on a N-anholonomic manifold V (in particular, on a tan-
gent bundle TM) with prescribed h-/v-splitting. Such a model is completely dened
on TV if there are certain geometric principles formulated for generation by F,
in a unique form, a triple of fundamental geometric objects (
F
N,
F
g,
F
D). Via
frame transforms, any such triple transforms into a general one (N, g, D), with
decompositions = D + Z, when a d-connection D, a LeviCivita connection
and a distorting tensor Z are uniquely determined by N and g. Inversely,
we can always introduce Finsler variables (with left up F-label) for any data
(N, g, D). For simplicity, we shall omit hereafter the label F that will not result in
ambiguities.
There are two general classes of Finsler geometries: (1) metric compatible, which
can be included into (or minimally extending) standard theories, when Dg = 0, and
(2) metric noncompatible (generating nonstandard models), when Dg = Q = 0.
For any given F, and/or g, there is a unique canonical d-connection

D (A.3) when
Dg = 0 and h- and v-torsions vanish but the general nonzero torsion

T is induced
by N and g. On spaces of odd dimension,

D can be transformed into the h-v
d-connection

D is determined by a couple of coecients

= (
L
a
bk
,

C
a
bc
) (30).
Via frame transforms, we can relate this d-connection to the Cartan d-connection
c
D = {
c
L
i
ik
,
c
C
a
bc
} (17) for Finsler geometry, which also denes a canonical almost
symplectic d-connection (see details in Refs. 18, 19, 10 and 15). We can work
1250072-19
S. I. Vacaru
equivalently on a nonholonomic space V with any

D,

D and/or
c
D in N-adapted
form.
The key issues for elaborating a Finsler generalization of Einstein gravity are
to introduce on V (in particular on TM) a metric g with pseudo-Euclidean sig-
nature and to decide what type of metric compatible d-connection D will be used
for postulating the eld (EinsteinFinsler) equations. For physically viable Finsler
gravity theories, any generalized Finsler fundamental geometric objects g and D
should contain as particular cases certain Einstein gravity solutions. We proved
some important results
10,22,26,33
that for

D and/or

D
c
D, the Einstein equations
can be integrated in very general forms. Imposing the zero torsion constraints
(22), when

D, we restrict the integral varieties to dene general solutions in
Einstein gravity and its higher dimension generalizations.
3.2.2. Principles of EinsteinFinsler relativity
Finsler congurations in GR: Finsler variables and the canonical d-connection
D =

Z (16), (similarly,

D and/or
c
D) can be introduced in GR if nonholo-
nomic 2+2 splitting are considered for a nonholonomic pseudo-Riemannian space-
time V.
10,15
All geometric and physical objects and fundamental equations can be
re-expressed in terms of

D and N-adapted variables. Such a formal Finsler gravity
satises all axioms introduced for the Einstein gravity theory. So, alternatively to
well-known tetradic, spinor, Ashtekar and other variables in GR, we can introduce
nonholonomic/Finsler variables.
Minimal Finsler extensions on TM of the standard model: The concept
of at Minkowski spacetime, with pseudo-Euclidean signatures, and postulates for
SR are related to the Maxwell electromagnetic eld theory. They were formulated
following MichelsonMorley type experiments with constant speed of light. The
most important symmetry is that of Lorentz (pseudo-rotation) and Poincare (with
translations) invariance with respect to certain linear group transforms.
Any possible contributions from QG result in nonlinear dispersions for light
rays of type (2) and nonlinear quadratic elements (1). To explain such physical
eects and elaborate generalized models of classical and quantum theories there
were considered various generalizations/restrictions of symmetries in SR
7,11,32,36
when the Minkowski metric
ij
= [1, 1, 1, 1] transform into a Finsler type g
ij
(y)
depending locally on velocity/momentum type co-ordinates y
i
. It is not clear from
general physical arguments why certain models of broken Lorentz invariance should
have priorities with respect to another ones. Perhaps, in a minimal way we can
say that
ij
g
ij
(y) is similar in some lines with generalizations of SR to GR,
ij
g
ij
(x), but in our case we may have an additional curvature determined
by bers of a co/tangent bundle, in general, by metrics of form g
ij
(x, y). Such a
metric should be a solution of the EinsteinFinsler equations and may possess some
nonholonomically deformed Lorentz symmetries.
1250072-20
Generalized equivalence principle: In Newtonian theory of gravity, the exper-
imental data show that the gravitational force on a body is proportional to its
inertial mass. This supports a fundamental idea that all bodies are inuenced by
gravity and, indeed, all bodies fall precisely the same way in gravitational elds.
Because motion is independent of the nature of the bodies, the paths of freely falling
bodies dene a preferred set of curves in spacetime just as in special relativity the
paths in spacetime of inertial bodies dene a preferred set of curves.
The world lines of freely falling bodies in a gravitational eld are simply the
geodesics of the (curved) spacetime metric. This suggests the possibility of ascribing
properties of the gravitational eld to the structure of spacetime itself. Because
nonlinear dispersions from Minkowski spacetime can be associated to metrics of type
g
ij
(y), and GR to metrics of type g
ij
(x), we can consider a generalized equivalence
principle on Finsler spacetimes with metrics of type
F
g
ij
(x, y). We may preserve the
ideas of universality of free fall and the universality of the gravitational redshift in
a Finsler type spacetime modeled by data (N, g, D). In such a locally anisotropic
spacetime, the paths of freely falling bodies are not usual geodesics but certain
nonlinear (semi-spray) ones which are dierent from auto-parallels of D (see details
on such geometries in Refs. 10, 22, 19 and 23).
v
Working with metric compatible d-connections completely determined by the
metric and N-connection structures, we can establish a one-to-one correspondence
between one type of preferred curves (semi-sprays) and auto-parallels of D. This
way we can encode equivalently the experimental (curvature deviation) data with
respect to both types of congruencies. In all important physical equations for a
Finsler gravitational and matter elds, the connection D (for canonical construc-
tions, it is used

D,

D and/or
c
D) is contained. Such a d-connection can be used
for constructing the Dirac, dAlambert and other important operators which allows
us to compute the light and particle propagation in a Finsler spacetime.
Generalized Mach principle: The Einstein gravity theory was formulated using
a second much less precise set of ideas which goes under the name of Machs prin-
ciple. In SR and in pre-relativity notions of spacetime, the geometric structure of
spacetime is given once and for all and is unaected by the material bodies that
may be present. In particular, the properties of inertial motion and nonrotating
are not inuenced by matter in the universe. Mach supposed that all matter in the
universe should contribute to the local denition of nonacceleration and non-
rotating. Einstein accepted this idea and was strongly motivated to formulate a
theory where, unlike SR, the structure of spacetime is inuenced by the presence of
matter. In GR, such purposes were achieved only partially. With respect to Finsler
v
For models of generalized Finsler spacetimes, it is important to study the geometry of semi-
spray congurations as N-connection generalizations of autoparallel and geodesic curves. In some
sense, semi-sprays characterize the N-connection eects into physical paths of test particles.
We recommend the interested readers to consult the respective sections in the mentioned review
papers and monographs.
1250072-21
S. I. Vacaru
gravity theories on (co)tangent bundles derived from quantum nonlinear dispersion
we can consider a generalized Mach principle that quantum energy/motion should
contribute to spacetime, i.e. the structure of spacetime is inuenced by the presence
of quantum world. This inuence is encoded both into the nonholonomic structure
and via coecients of (N, g, D) into energymomentum tensors for matter elds
imbedded self-consistently in spacetime aether with moving co-ordinates y
a
.
EinsteinFinsler spacetimes and gravitational equations: New theories
of locally anisotropic spacetime and gravitation state the following: The intrinsic,
observer-independent, properties of Finsler spacetime are described by a Finsler
generating functions which canonically determine the N-connection, d-metric and
d-connection fundamental geometric objects in a metric compatible form as in GR
but for N-adapted constructions. We dene a Finsler gravity model and its fun-
damental gravitational equations on a N-anholonomic manifold V, including GR
and SR (as certain particular classes of solutions) following the same principles as
in Einstein theory but in N-adapted form for a xed canonical metric compati-
ble d-connection (a Finsler d-connection) which is dierent from the LeviCivita
connection.
Principle of general covariance: In GR, this is a natural consequence from
the fact that spacetime models are constructed on (pseudo) Riemannian mani-
folds. So, the geometric and physical constructions do not depend on frames of
reference (observers) and co-ordinate transforms. In denition of Finsler geometry
models the concept of manifold is also involved (in certain approaches such man-
ifolds are tangent/vector bundle spaces). So, the principles of general covariance
has to be extended on V, or TM. We can introduce certain preferred systems of
reference and adapted co-ordinate transforms when a xed h-v-decomposition is
preserved/distinguished but this is a property of some particular classes of solu-
tions of the EinsteinFinsler equations. In general, we cannot distinguish between
triples of data (
F
N,
F
g,
F
D) and (N, g, D). We can use any parametrizations
of Finsler data which are necessary for certain construction in a model of classi-
cal or QG. In an extended Finsler spaces principle of generalized covariance (for
instance, for the canonical d-connection), there are included distortion relations of
type

D =
Z (16). So, we can describe geometrical and physical models equiva-

lently both in terms of

D and because such connections are dened by the same
metric structure.
The equations of motion and conservation laws: The conservation law
i
T
ij
= 0 is a consequence of the Bianchi relations and involve the idea that
in GR the Einsteins equations alone actually implies the geodesic hypothesis (that
the world lines of test bodies are geodesics of the spacetime metric). Note however,
that bodies which are large enough to feel the tidal forces of the gravitational
eld will deviate from geodesic motion. Such deviations may be caused by certain
nonholonomic constraints on the dynamics of gravitational elds. The equations of
motion of such bodies in GR also can be found from the condition
a
T
ab
= 0.
1250072-22
For a Finsler d-connection D, even it is metric compatible, D
= 0, which is
a consequence of nonsymmetry of the Ricci and Einstein d-tensors, see explanations
for formula (15) and generalized Bianchi identities. Such a property is also related
to nonholonomic constraints on the dynamics of Finsler gravitational elds. It is
not surprising that the covariant divergence of source does not vanish even for
D,

D and/or
c
D. Using distorting relations of type

D =

Z (16), we can
always compute

D
from
i
T
ij
for matter elds moving in a canonical Finsler
spacetime following principles minimally generalizing those for GR as we explained
above. In this case, the conservation law are more sophisticated by nonholonomic
constraint but nevertheless it is possible to compute eective nonholonomic tidal
forces of locally anisotropic gravitational elds when auto-parallels of

D deviated
from nonlinear geodesic (semi-spray) congurations.
Axiomatics for the EinsteinFinsler gravity: A constructive-axiomatic
approach to GR was proposed in 1964 by Ehlers, Pirani and Schild
45
(the so-called
EPS axioms). In a series of publications in the early 1970s and further develop-
ments (see original results and references in Refs. 46, 47, 35 and 48), it was elab-
orated the concept of EPS spacetime as a physically motivated geometric model
of spacetime geometry. That axiomatic approach led to a common belief that the
underlying geometry of the spacetime can be only pseudo-Riemannian which lead
to the paradigmatic concept of Lorentzian four-manifold in GR.
An axiomatic approach to Finsler gravity theory was proposed in Ref. 49; a
minimal set of axioms for Finsler geometry was also formulated.
18
We consider that
it is not possible to elaborate a general EPS type system for all types of Finsler
gravity theories. For the EinsteinFinsler spaces, the EPS axioms can be extended
on TM when

D/
c
D are used for denition of auto-parallels and light propagation
on nonholonomic manifolds.
3.2.3. On physical meaning of velocity/momentum co-ordinates
We can consider the GR theory as a branch of modern mathematics when the
physical theories are geometrized on a Lorentz manifold V. The concepts of tan-
gent Lorentz bundle, TV, and its dual, T
V, are well-dened. For corresponding

local co-ordinates (x, y) = (x
i
, y
a
) and (x, p) = (x
i
, p
a
), the values y
a
are called
velocity type co-ordinates and p
a
are dual, or momentum type co-ordinates.
Geometrically, the gravity and cosmological and physical models on TV, or T
V,
can be elaborated as higher dimension ones for metric-ane
22
or (super) Finsler
superstring theories,
20
with velocity/momentum extra-dimensional co-ordinates.
Such theories cannot be compactied on y
a
because there are a constant velocity
of light in vacuum and, in general, a nontrivial N-connection structure. There are
important various o-diagonal trapping/warping eects for Finsler branes.
50,51
The
corresponding metrics are generic o-diagonal, the N- and d-connections are with
nontrival torsion, or/and with nonzero nonmetricity elds, depending on veloc-
ity/momentum type variables. Using N-adapted lifts and corresponding covariant
1250072-23
S. I. Vacaru
dierential and variational calculus, we can extend geometrically any physical the-
ory on V to TV and/or T
V. We can consider such theories as a simple framework

for encoding properties of at quantum spacetime.
Nevertheless, only a formal geometric approach does not allow us to provide
a well-dened physical meaning to y
a
and/or p
a
. Such values can be naturally
introduced in geometric mechanics and eective (commutative and noncommuta-
tive, or (super) symmetric) theories of gravity,
19,22
but their physical interpretation
depends on explicit form of models we elaborate. Let us analyze and discuss in brief
several important works developing theories with y
a
and/or p
a
.
Nonlinear deformed relativity (or doubly special relativity, DSR; see
Refs. 2729 and references therein): This class of theories is constructed via
nonlinear realizations of the Lorentz group when modied Lorentz transformations
reduce to the usual ones at low energies. There are considered deformed dispersion
relations and an invariant E
P
as a borderline between classical and QG. For a
number of historical and phenomenological reasons (the theory is motivated by
cosmic ray kinematics) the nonlinear relativity was rst studied on momentum
space when recovering the position space is highly nontrivial.
Considering a dual to nonlinear realizations of relativity in momentum space,
there were constructed energy-dependent metrics, connection and curvature for
a simple modication to Einsteins equations. Several counterparts to the cos-
mological metrics where found and shown how cosmologies based upon energy-
depending theory of gravity may solve the horizon problem. There were some
solutions explored to this theory of gravity, namely the cosmological, black hole
and weak eld solutions. Such results may have important implications for black
hole thermodynamics.
It should be noted that dierent theories of gravity are derived for dierent
realizations of position space. If the Lorentz group is nonlinearly realized in the
position space, gravity is induced gauging a symmetry which is nonlinearly realized.
It is possible to consider noncommutative positions and/or momentum spaces.
Quantum-gravitational uctuations in the spacetime and D-
brane/-particle foam
3,4,30,31
: The main idea was that quantum-gravitational
uctuations in the spacetime background induce nontrivial optical properties of the
vacuum. That include diusion and stochastic properties and consequent uncertain-
ties in the arrival times of photons. Such an approach can be motivated within a
Liouville string formulation of QG and suggests a frequency-dependent refractive
index of the vacuum for particles and quantum elds. An explicit realization can be
constructed by treating photon propagation through quantum excitations and, in
general, anisotropic D-brane uctuations in the spacetime foam. This way we can
describe string eects that lead to stochastic uctuations in couplings and hence in
the velocity of light.
1250072-24
Within the context of supersymmetric spacetime (D-particle) foam in
string/brane-theory, it is possible to construct models of Finslerinduced cosmol-
ogy and to study possible implications for (thermal) dark matter abundances. In
such approaches, there are elaborated microscopic models of dynamical spacetime,
where Finsler geometries arise naturally. For instance, we can consider eects of
recoil of D-particles related to a back-reaction on the spacetime metric of Finsler
type which is stochastic. The induced Finsler-type metric distortions depend (addi-
tionally to the spacetime co-ordinates) also on the pertinent momentum transfer
which provides an explicit physical interpretation of such co-ordinates.
Classical and QG models with N-connection on TV and/or T
V
5153
:
There is a recent interest in physics beyond the standard model which can incor-
porate Lorentz symmetry violations, accelerating universe and dark energy/matter
eects in a Finsler setting which was discussed in above subections (see also Refs. 54,
32, 7, 11, 55, 12, 8 and 9). The Finsler metrics are functions not only of the spacetime
co-ordinates but also of the tangent vectors (momenta) at points of the curved man-
ifolds. Certain ansatz for toy models were parametrized in diagonal form,
2730
or
with some examples of o-diagonal stochastic metrics.
4,31
The fundamental issues of
Finsler classical and QG and cosmology related to N-connection and nonholonomic
structures were not studied in above mentioned references. In variables (x
i
, p
a
), and
similar ones on higher-order (co)tangent bundles, the problem of constructing N-
adapted classical and QG theories was studied in Refs. 52, 22 and 53. In such works,
the velocity/momentum type co-ordinates can be associated to (in general, higher-
order) spinor and/or almost Kaehler variables which can be used, for instance, for
deformation quantization of such theories.
Finally we note that geometrically all above mentioned model depending
constructions can be formalized using the concept of nonholonomic tangent
bundle/manifold which, for nontrivial limits to standard GR, should involve non-
holonomic deformations of Lorentz manifolds. A number of important physical
issues on fundamental property of cosmological models for such locally anisotropic
gravity theories (like viable cosmological models with acceleration, diagonal and
generic o-diagonal Finsler evolution, etc.) can be studied following geometric and
analytic methods of constructing exact solutions, symmetries of such solutions
and evolution scenarios.
4. Accelerating Cosmology as Finsler Evolution
In this section, we show that the acceleration expansion of the present matter-
dominated universe may be generated along with the evolution of Finsler space in
velocity type dimensions. Two examples of exact o-diagonal solutions associated
with cosmological evolution scenarios will be constructed. We prove that solitonic
nonholonomic deformations induced by velocity type variables modify scenarios of
acceleration in real universe.
1250072-25
S. I. Vacaru
4.1. Diagonal accelerating Finsler universes
We consider a prime metric of a (n+1+3)-dimensional spacetime, with n = 4 and
m = 1 + 3, with time like co-ordinate y
5
= t,
0
g =
1
dx
1
dx
1
+
h
a
2
(t)
1
h
k(
h
r)
2
d
h
r d
h
r +
h
a
2
(t)(
h
r)
2
d
h
d
h
+
h
a
2
(t) (
h
r)
2
sin
2 h
d
h
d
h
dt dt +
v
a
2
(t)
1
v
k(
v
r)
2
d
v
r d
v
r
+
v
a
2
(t)(
v
r)
2
d
v
d
v
+
v
a
2
(t) (
v
r)
2
sin
2 v
d
v
d
v
. (37)
This diagonal ansatz is considered for a simple cosmological model, with zero
N-connection coecients, on a tangent bundle to a 4-d (pseudo) Riemannian man-
ifold, when

D = .
w
We study a toy model on TV for a metric with both the
h- and v-parts of FRW type, i.e. spherical symmetries on h- and v-co-ordinates.
The constructions can be involved in a class of solutions for nonlinear deformed
gravity (in some sense, with double FRW position and phase cosmology).
To generate such solutions we have to chose a source

on TV for the gen-

eralized Einstein equation (21), using N-adapted lifts from hTV to TV, when a
perfect uid on the horizontal part possess certain vertical anisotropies resulting
in acceleration of observable h-subspace. We do not have experimental data which
would suggest how a perfect v-uid could move in such extra dimensions (here
we note that in the diagonal case, the N-connection is trivial). Nevertheless, we
can approximate the dynamics of such locally anisotropic distributions of matter
to be with eective density function
v
and pressure
v
p stated as anisotropically
polarized cosmological constants in a 8-d bulk. It should be noted that the physical
interpretation of v-co-ordinates depend on the type locally anisotropic theory we
consider for our cosmological model (as we explained in Sec. 3.2.3). Geometrically,
such constructions are for generic diagonal solutions with double spherical sym-
metry of (18)(20) which will be used as prime metrics for generating more realistic
o-diagonal locally anisotropic models, see next section and discussions for metrics
(41) and (45).
w
The local co-ordinates are u
= (x
i
, y
a
), for i, j, . . . = 1, 2, 3, 4 and a, b, . . . = 5, 6, 7, 8, (the time
like co-ordinate t is considered as the rst ber co-ordinate) and the coecients of
0
g
=
diag[
0
g
i
,
0
h
a
] are, for spherical h-co-ordinates: x
1
= x
1
, x
2
=
h
r, x
3
=
h
, x
4
=
h
and, for
v-co-ordinates, y
5
= t, y
6
=
v
r, y
7
=
v
, y
8
=
v
;
0
g
1
=
1
= 1,
0
h
5
= 1,
0
g
2
=
h
a
2
(t)
1
h
k(
h
r)
2
,
0
g
3
=
h
a
2
(t)(
h
r)
2
,
0
g
4
=
h
a
2
(t) (
h
r)
2
sin
2 h
,
0
h
6
=
v
a
2
(t)
1
v
k(
v
r)
2
,
0
h
7
=
v
a
2
(t)(
v
r)
2
,
0
h
8
=
v
a
2
(t) (
v
r)
2
sin
2 v
.
1250072-26
4.1.1. Diagonal cosmological equations for pseudo-Finsler metrics
The metric (24) describes two types (conventional horizontal and vertical ones) of
evolutions with time variable t of two universes with respective constant curvatures
h
k and
v
k. To derive cosmological solutions in a most simple form is convenient
to consider in the h-subspace a radial co-ordinate 0 <
h
r < 1 taken for the light
velocity c = 1. The coecients
0
h
a
dene a usual FRW type metric in the v-
subspace. The values
h
a
2
(t) and
v
a
2
(t) are respective h- and v-scale factors. In
such models, the h-co-ordinates are dimensionless (we can introduce a standard
dimension, for instance, by multiplying on Planck length) and the v-co-ordinates
are usual ones, with dimension of length.
Assuming that the matter content in this pseudo-Finsler spacetime is taken to
be a perfect uid, we can write the Einstein equations (26)(29) as
4
v
a
v
a
h
a
h
a
+ 2
h
a
h
a
2
+
h
k
(
h
a)
2
v
a
v
a
2
+
v
k
(
v
a)
2
=
8
3
G,
4
h
a
h
a
+ 2
v
a
v
a
+ 6
h
a
h
a
2
+
h
k
(
h
a)
2
v
a
v
a
2
+
v
k
(
v
a)
2
= 8G
v
p,
h
a
h
a
+
v
a
v
a
+ 2
h
a
h
a
2
+
h
k
(
h
a)
2
v
a
v
a
2
+
v
k
(
v
a)
2
=
8
3
G
h
p.
(38)
In the above formulas, the right dot means derivative on time co-ordinate t and
G and are respectively the formal gravitational constant and the energy density
in the total (tangent) space. The values
h
p and
v
p are, correspondingly, the pres-
sures in the h- and v-spaces. We assume simple equations of matter states of type
h
p =
h
h
and
v
p =
v
v
for some constant state parameters
h
and
v
.
x
The
conservation law
= 0 for T
= diag[
h
,
h
p, . . . ;
v
,
v
p ] gives rise to

h
a
4(1+
h
)
v
a
3(1+
v
)
. For simplicity, we may assume
h
k = 0 and study the
evolution of the scale factors
h
a(t) and
v
a(t) and approximations in (38).
4.1.2. Diagonal scale evolution and velocity type dimensions
In general, a Finsler gravity dynamics is with generic o-diagonal metrics and
generalized connections. Such nonlinear systems may result in non-perturbative
eects and instability even for small o-diagonal metric terms.
x
From a formal point of view, we can construct on tangent bundles perfect uid models with
formal dierent h- and v-pressures, in N-adapted form as we discussed in Ref. 22 (we omit in this
work such details and send the reader to a paper on anisotropic inationary model by Vacaru
and Gonta, in that collection of papers).
1250072-27
S. I. Vacaru
Radiation-dominated diagonal Finsler universe: We dene such an universe
following conditions
h
p = 0 and
v
p =
1
3
v
, when
h
a = const. is accepted as a
solution. For such congurations, the third equation in (38) is a consequence of the
rst two ones when
v
a(t) is a solution of equations:
v
a
v
a
2
+
v
k
(
v
a)
2
=
8
3
G, 2
v
a
v
a
+
v
a
v
a
2
+
v
k
(
v
a)
2
=
8
3
G.
The source
8
3
G of such equations is determined by generalized gravitational con-
stant G and matter density in total spacetime. By straightforward computations,
we can show that the constant
h
a-solution is stable under small perturbations of
scale factors
h
a(t) and
v
a(t). This means that we can retrieve the ordinary evolu-
tion of radiation-dominated Finsler universe with a total spacetime model. Here we
note that for a matter-dominated Finsler conguration with
h
p =
v
p = 0 there is
not a solution with
h
a = const. unless = 0.
Matter-dominated diagonal Finsler universe: There are solutions as in the
standard FRW cosmology (in our case, for the v-part) with
h
a = const., when the
matter in the velocity space provides negative pressure
h
p =
1
2
, when
v
p = 0.
Such conditions may be realistic if we associate point like nonrelativistic particles in
v-space certain extended objects (let say, strings) with additional velocity variables
when the pressure is provided in such a strange manner. The FriedmanFinsler
equations (38) transform into (
v
a
v
a
)
2
+
v
k
(
v
a)
2
=
8
3
G, 2
v
a
v
a
+(
v
a
v
a
)
2
+
v
k
(
v
a)
2
= 0,
which allows us to nd general solutions for
v
a(t).
Dierent extension rates in h- and v-subspaces: For simplicity, we can
assume
h
k =
v
k = 0 and that some constants
h
and
v
determine
h
p =
h
and
v
p =
v
v
, i.e. the equations of states in a matter like dominated Finsler
universe. The dierence between the v- and h-expansion rates is expressed
v
h
(t) :=
(13
v
+2
h
)
v
a
v
a
[1+3
v
4
h
]
h
a
h
a

1
(
v
a)
3
(
h
a)
4

1
Vol
3+4
, where Vol
3+4
is
the volume of a (3+4)-dimensional space like total pseudo-Finsler subspace if signa-
ture
1
= 1 in (38). It follows from this formula that the dierence
v
h
(t) decreases
(grows) as the volume Vol
3+4
grows (decreases). Here we note that (
h
a)
4
has a
limit corresponding to the maximal velocity of light.
Similarly, it is possible to consider the dierence between v- and h-expansions
v
h
(t) for a radiation-dominated Finsler universe with
h
p =
1
3
and
v
p = 0,
v
h
(t) :=
2
h
a
h
a

1
(
v
a)
3
(
h
a)
4

1
Vol
3+4
. We conclude that if Vol
3+4
is growing, the expansion
rate of the h-spaces drops to zero. So, the constant
h
a solution is stable for the
radiation dominated Finsler universe.
It is possible to consider a more general matter dominated Finsler universe with
h
=
v
=
1
3
when
v
h
(t)
v
a
v
a

h
a
h
a

1
(
v
a)
3
(
h
a)
4

1
Vol
3+4
.
If the total volume Vol
3+4
is growing, the expansion rates of the h- and v-spaces
tend to approach each other, i.e. the limited h-volume, because of nite speed of
1250072-28
light, limits the three-space in the v-part. If for an inverse decreasing of Vol
3+4
,
with one expanding and another collapsing subspaces, then |
v
a
/
v
a|, or |
h
a
/
h
a|,
becomes large and larger. This results in an accelerating expansion. For collapsing
h-space with velocity types co-ordinates we can induce an accelerating expansion of
our inverse modelled by this pseudo-Finsler model as the v-subspace. We analyze
below more details on such models of Finsler-acceleration.
4.1.3. Accelerating diagonal expansion with Finsler evolution
We explore analytically the possibility to generating Finsler-type acceler-
ating expansions via evolution of h-space with velocity co-ordinates. For
simplicity, we consider
h
k =
v
k =0 and trivial equations of states with
h
p =
v
p =0.
y
For such conditions, the last two equations in (38) became
v
H
+
5
2
(
v
H)
2
+2
v
H
h
H (
h
H)
2
=0,
h
H
1
2
(
v
H)
2
+
v
H
h
H+3(
h
H)
2
= 0, where the
respective eective Hubble h- and v-constants are
h
H :=
h
a
/
h
a and
v
H :=
v
a
/
v
a. These equations impose the corresponding conditions for accelerating
(
v
a
/
v
a > 0), or decelerating (
v
a
/
v
a < 0) of our three-dimensional v-subspace,
acceleration:
h
H > (1+
5/2)
h
H :=
+
H
h
H, or
h
H > (1
5/2)
h
H :=

H
h
H;
deceleration:

H
v
H <
h
H <
+
H
v
H. To investigate the correlation between
h- and v-subspaces is useful to introduce the fraction-function (t) :=
h
H/
v
H and
see the behavior of d/dt for dierent values of and some critical values of this
function, which for our dimensions n = 4 and m = 1 + 3 are dened: attracting:
att
H := 1 + 1/
2; repelling:
rep
H := 1 1/
2. It is always satised the condi-

tion
att
H <

H <
rep
H < 0 < 1 <
+
H, i.e. there are two attractors determined
by =
att
H and = 1 and one repeller for =
rep
H. One follows the conditions
> 0, for <

att
H,
rep
H < < 1;
< 0, for
att
H < <
rep
H, > 1.
For Finsler universes, there are four kinds of evolution processes depending of
a initial value =

:
acceleration and, then, deceleration,

>
+
H;
always deceleration,
rep
H <

<
+
H;
deceleration and, then acceleration,

H <

<
rep
H;
always acceleration,

<

H.
(39)
A realistic condition for our universe is the third condition, when

H <

<
rep
H.
Such a scenario states that initially the Finsler universe is in the region (
H,
rep
H)
when the h-space collapses and our three dimension space in the v-part decelerates.
As passes

H in the collapsing process of velocity h-co-ordinates, our real
3-d space begins to accelerate.
y
Models with nonzero
h
k and/or {}
v
k oer a number of interesting possibilities. In the next
subsections, we shall investigate examples with nontrivial N-connection and Riemannian and scalar
curvatures.
1250072-29
S. I. Vacaru
4.2. O-diagonal anisotropic Finsler acceleration
More realistic Finsler-type cosmological models can be elaborated for generic o-
diagonal metrics and with nontrivial N-connection.
4.2.1. Examples of o-diagonal cosmological solutions
We construct in explicit form two classes of such solutions dening certain models
of four-dimensional, 4-d and 8-d Finsler spacetimes.
A pseudo-Finsler 4-d o-diagonal toy cosmology: Let us consider:
g =
h
a
2
(t)d
h
r d
h
r dt dt +
h
a
2
(t)(
h
r)
2
d
h
d
h
+
h
a
2
(t)(
h
r)
2
sin
2 h
d
h
d
h
, (40)
which is contained as a particular case of 8-d ansatz (37), when
1
= 0,
v
a = 0 an,
for simplicity,
h
k =
v
k = 0. We use this metric for a prime cosmological model in
variables u
b
= (
h
r, t,
h
,
h
), with x
b
i
= (
h
r, t) and y
ba
= (
h
,
h
), for
i,
j, . . . = 2, 3
and a,
b, . . . = 4, 5 (such a model describes evolution in time t of a h-subspace for

certain conditions, and sources, analyzed in Sec. 4.1). An o-diagonal anisotropic
dynamics in the space of velocities can be modeled by nonholonomic deformations
with -polarizations
b
i
=
b
i
(x
b
k
) and
ba
=
ba
(x
b
k
,
h
) and N-connection coecients
N
b
4
b
i
= w
b
i
(x
b
k
,
h
) and N
b
5
b
i
= n
b
i
(x
b
k
,
h
). For the prime metric g
b
2
=
h
a
2
(t), g
b
3
=
1,
h
b
4
=
h
a
2
(t)(
h
r)
2
,
h
b
5
=
h
a
2
(t)(
h
r)
2
sin
2 h
, we dene g = [ g
b
i
,
h
ba
,

N
ba
b
j
] g =
[g
b
i
=
b
i
g
b
i
, h
ba
=
ba
h
ba
, N
ba
b
i
], to a metric
4d
g = g
b
i
d x
b
i
dx
b
i
+h
ba
(dy
ba
+N
ba
b
i
dx
b
i
) (dy
ba
+N
ba
b
i
dx
b
i
), (41)
constrained to be a cosmological solution of Eqs. (31)(34) for

D. For simplic-
ity, we consider a source with constant coecients
b
b
= diag[
b
;
2
=
3
= const.;
4
=
5
= const.] transforming for a diagonal limit into T
=
diag[
h
,
h
p, . . . ;
v
,
v
p ] used for generating a metric (38).
The coecients of such a new solution (41) are of type (36) generated from (40)
by polarization functions and N-connection coecients
z
:
2
= e
(
h
r,t) h
a
2
(t),
3
= e
(
h
r,t)
for
=
2
,
4
= [f
(
h
r, t,
h
)]
2
|(
h
r, t,
h
)|,
for = 1

4
8
d
h
f
(
h
r, t,
h
) [f(
h
r, t,
h
)
0
f(
h
r, t)],
z
To simplify formulas, we chose corresponding parametrizations for generating/integration
functions.
1250072-30
5
= [f(
h
r, t,
h
)
0
f(
h
r, t)]
2
,
w
b
j
=
0
w
b
j
(
h
r, t) exp
2
4
A
vv
1
dv
1
0
dv
1
4
B
j
vv
1
exp
v
1
0
2
4
A
vv
1
dv
1
,
n
b
i
=
0
n
b
i
(
h
r, t) +
d
h

4
h
a
2
(t)(
h
r)K
b
i
, (42)
where the coecients of type (35) are computed for polarization functions,
A =
4
2
4
+

5
2
5
, B
b
k
=

5
2
5
b
k
g
b
2
2g
b
2

b
k
g
b
3
2g
b
3
b
k
A,
K
2
=
1
2
2
g
3
h
5
+
g
3
g
3
h
5
, K
3
=
1
2
3
g
2
h
4
3
g
3
h
5
.
(43)
In formulas (42) and (43), the partial derivatives are written in brief in the form
4
=
4
/
h
, g
3
= g
3
/
h
r, g
3
= g
3
/t and
0
f(
h
r, t),
0
w
b
j
(
h
r, t),
0
n
b
i
(
h
r, t) are
integration functions to be determined by xing some boundary/initial conditions
in the space of velocities.
Putting together the above coecients, we nd the 4-d metric
g = e
(
h
r,t)
(d
h
r d
h
r dt dt) + [f
(
h
r, t,
h
)]
2
|(
h
r, t,
h
)|
h
a
2
(t)(
h
r)
2
h

h
+ [f(
h
r, t,
h
)
0
f(
h
r, t)]
2 h
a
2
(t) (
h
r)
2
sin
2 h
h

h
,
h
= d
h
+w
b
2
(
h
r, t,
h
)d
h
r +w
b
3
(
h
r, t,
h
)d
h
r,
h
= d
h
+n
b
2
(
h
r, t,
h
)d
h
r +n
b
3
(
h
r, t,
h
)d
h
r,
with the coecients dened by data (42). Such a metric dened an o-diagonal
Finsler inhomogeneous model in the h-subspace. In our real universe it may
contribute via a nontrivial e
(
h
r,t)
before time like dt; such a solution should be
imbedded into a 8-d Finsler spacetime.
A class of inhomogeneous o-diagonal 8-d Finsler cosmologies: Following
the geometric method of constructing exact solutions in extra dimensional space-
time,
33
we can generalize the metric (41) with coecients (42) to generate cosmo-
logical solutions for a total 8-d Finsler spacetime. We consider a source (B.1) with
constant coecients modeling on h- and v-subspaces perfect uid matter/radiation
states. The 8-d ansatz is
8d
g =
1
dx
1
dx
1
+
b
i
g
b
i
d x
b
i
dx
b
i
+
4
h
4
(dy
4
+w
b
i
dx
b
i
) (dy
4
+w
b
i
dx
b
i
)
+h
5
(dy
5
+w
(u
,
1
v)du
) (dy
5
+w
(u
,
1
v)du
)
1250072-31
S. I. Vacaru
+h
6
(dy
6
+n
(u
,
1
v)du
) (dy
6
+n
(u
,
1
v)du
)
+h
7
(dy
7
+w 1
(u
1
,
2
v)du
1
) (dy
7
+w 1
(u
1
,
2
v)du
1
)
+h
8
(dy
8
+n 1
(u
1
,
2
v)du
1
) (dy
8
+n 1
(u
1
,
2
v)du
1
), (44)
where co-ordinates and respective indices are parametrized u
= (x
1
, x
b
i
=
(
h
r, t), x
4
= y
4
=
h
); u
1
= (u
, y
1
a
= (y
5
=
1
v =
v
r, y
6
=
h
)); u
2
=
(u
2
, y
2
a
= (y
7
=
2
v =
v
, y
8
=
v
)) and h
6
=
6
h
6
+h
6
(for h
6
known for given
h
6
and
6
h
6
), n
(u
,
1
v) = (n
b
i
(x
b
k
,
h
); n
(u
,
1
v), if > 3) when the values
b
i
g
b
i
,
4
h
4
,
6
h
6
being former
5
h
5
in (42), w
b
i
and n
b
i
are given by coecients of
metrics (41) and (40).
aa
From the class of general solutions, we can extract a subclass of 3-d solitonic
congurations = (t,
h
,
v
r) from the hsubspace, depending on time and velocity
type co-ordinates, inducing small perturbations in the vsubspace. Such anisotropic
on velocities 8d metrics are written:
sol
g =
1
dx
1
dx
1
+e
(
h
r,t) h
a
2
(t) d
h
r d
h
r
+
4
(
h
r, t,
h
)
h
a
2
(t)(
h
r)
2
h
(
h
r, t,
h
)
h
(
h
r, t,
h
)
+
6
[]
5
(
h
r, t,
h
)
h
a
2
(t)(
h
r)
2
sin
2 h

h
[]
h
[]
e
(
h
r,t)
dt dt + (1 +
5
[])
v
a
2
(t)
v
r[]
v
r[]
+
v
a
2
(t)(
v
r)
2
d
v
d
v
+
v
a
2
(t)(
v
r)
2
sin
2
(
v
) d
v
d
v
, (45)
where
h
(
h
r, t,
h
) = d
h
+ w
2
(
h
r, t,
h
)d
h
r +w
3
(
h
r, t,
h
)dt,
v
r[] = d
v
r + w
3
[]dt + w
4
[]d
h
,
h
[] = d
h
+ n
3
[]dt + n
4
[]d
h
,
for values e
,
4
, w
2
, w
3
and
5
determined by formulas (42) and the coecients
depending functionally on [], for a 3-d solitonic function = (t,
h
,
v
r) (for
instance, being a solitonic solution as we computed the end of Appendix B). Such
3-d solitons were considered in our works on propagation of black holes in extra
dimensional spacetimes and on local anisotropic black holes in noncommutative
gravity.
21,41,56
Solitonic congurations can be stable and propagate from the space
of velocities into real universe for various models of Finsler cosmology.
For 0, the solutions (45) transform into the 8-d metric (37) with possible
nonholonomic generalizations containing solutions of type (41), (42).
aa
In standard cosmological models based on GR, it is also possible to dene inhomogeneous and
anisotropic congurations. For such cosmological spacetimes, the metrics do not depend explicitly
on ber/velocity type variables. Finsler anisotropic/inhomogeneous constructions are dened by
more complex diagonal and/or o-diagonal metrics on spacetimes with tangent bundle structure.
1250072-32
4.2.2. Solitonic N-connections and anisotropic acceleration
The pseudo-Finsler cosmological model described by (45) is generically o-
diagonal. The solitonic deformation contributes both to diagonal and o-
diagonal terms of metric. We can x a nonholonomic(co) frame of reference, e
=
(dx
1
, d
h
r,
h
,
h
, dt,
v
r, d
v
, d
v
), for an observer in a point (
v
r
0
,
v
0
,
v
0
),
for simplicity, considering that the velocity space is with
h
r = 0,
h
with
one anisotropic velocity
h
. There are two eective scaling parameters
h
a() =
(1 +())
h
a();
v
a() = (1 +
5
[()])
v
a(), when we approximate e
(
h
r
0
,)
=
1 + () with the solitonic function () taken for a redened time like variable
(t), when d = e
(0,t)
dt.
Let us introduce
h

H
:=
h
a
/
h
a =
h
H +
, for
h
H =
h
a
/
h
a,
v

H
:=
v
a
/
v
a =
v
H +
5
, for
v
H =
h
a
/
h
a, where
= /. The conditions for

acceleration, in our case modied by a solitonic function both for o-diagonal
and diagonal terms of metric, are redened in the form
h
H
h

H,
v
H
v

H
and /t . This introduces additional, solitonic type, correlations between h-
and v-subspaces via an additional nonholonomic deformation of fraction-function,
() :=
h

H/
v

H () + (
5
)
. Conclusions about Finsler-solitonic o-

diagonal acceleration, or deceleration, should be drawn from behavior of func-
tions
() and (). Respectively, the solitonic versions of Eq. (39) are
>
0, for <
att
H,
rep
H < < 1;
< 0, for
att
H < <
rep
H, > 1, and (the four
types of evolution processes depend on a initial value =

)
acceleration and, then, deceleration,

>
+
H;
always deceleration,
rep
H <

<
+
H;
deceleration and, then acceleration,

H <

<
rep
H;
always acceleration,

<

H.
A solitonically modied directly observable universe is

H <

<
rep
H. Such a
condition is very sensitive with respect to possible (o-) diagonal perturbations
from the space of velocities. This follows from the facts that the conditions of accel-
eration for and = + (
5
)
are, in general, dierent. Small modications

proportional to (
5
)
may transfer, for instance, an accelerating conguration

into decelerating, and inversely.
In this section we have investigated the scenario of producing the accelerating
expansion of the present universe via evolving small velocity type diagonal and o-
diagonal nonholonomic deformations of Finsler metrics. For a radiaton-dominated
cosmological model, such as the model of our early universe, we obtain stable cong-
urations with static velocity type co-ordinates. In this case, the existence of Finsler
type velocity co-ordinates may have no signicant inuence on 3-d observable space.
Here we note that on the contrary, diagonal solutions with static extra dimensions
does not exist for the present matter-dominated cosmologies.
There are four classes of evolution for the matter-dominated cases, as we derived
from our quantitative analysis of both types diagonal and o-diagonal solutions.
1250072-33
S. I. Vacaru
Cosmological models that decelerate rst and then accelerate are included into
the schemes. Therefore, the accelerating Finsler expansion of the present universe
may be described in our locally anisotropic scenario. We note that small solitonic
type deformations from the velocity type subspace may modify substantially the
character of acceleration of universe and conditions of stability or instability.
Let us discuss several important properties of locally anisotropic (Finsler
type) cosmological solutions constructed in this section. There are substantial
dierences if we compare our results with those obtained for diagonal congu-
rations in Refs. 2729. Our cosmological spacetimes are, in general, with non-
trivial N-connection structure, i.e. dene nontrivial Finsler cosmologies which
results in anisotropically accelerating universes. For o-diagonal Finsler cos-
mological solutions, we can dene commutative and/or noncommutative black
hole/ellipsoid/wormhole (or extra dimensional o-diagonal cosmological) congu-
rations as in Refs. 21, 56 and 34. Such exact solutions can be constructed in explicit
form for various models of trapping/warping of Finsler branes and anisotropies of
HoravaLifshitzFinsler type.
41,50,51
Finally, we note that o-diagonal metrics of type (41) and (45) posses certain
similarities with metrics (3.10) and (4.18) constructed in Ref. 31 for a dierent
type of QG phenomenology and accelerating cosmology derived for stochastic D-
particle foam etc. Even the string-brane stochastic models are constructed from
another fundamental mathematic and physical principles than those considered
in this work, for metric compatible Finsler gravity models, they can be unied
in low energy limits using generic o-diagonal stochastic solutions of generalized
Einstein equations as in Ref. 57. In both cases of stochasic D-particle foam and exact
solutions with stochastic generating function the cosmological aspects of theories
are related to anisotoropic modications of the Boltzmann equation (to get results
similar to Refs. 4 and 31 we should consider solutions from Ref. 57 when the v-co-
ordinates are considered for distorted metric with momentum transfer). Solutions of
the Boltzmann equations and corresponding generating functions or foam-modied
thermal dark matter relic abundances can be connected to standard cosmology in
certain trivial N-connection limits, and absence of the foam.
5. Concluding Remarks
To avoid repetition, in this concluding section we do not attempt to summarize all
of the issues and application we discussed. This is because as, in most cases, the
preliminary insight gained and perspectives cannot be summarized in a sentence or
two and this would not be very helpful to readers. However, we have encountered
two key issues: (1) The EinsteinFinsler gravity can be formulated following the
same principles as GR but on certain nonholonomic bundle/manifold spaces and
corresponding generalized Finsler connections (which are also uniquely dened by
the coecients of metric tensor in a metric compatible form). (2) Analyzing possible
implications of QG and related Lorentz violations in early universe and present day
1250072-34
cosmology, we derived very easily that the dynamics (in general, with nonholonomic
constraints) in the space of velocities contributes substantially to stability and
acceleration/deceleration stadia of cosmological models.
The main purpose of this work was to show how the main postulates for the
GR theory can be extended on nonholonomic tangent bundles/manifolds. It was
provided a self-consistent scheme for formulating Finsler gravity models and fun-
damental physical equations in a form most close to standard particle physics.
Applying the anholonomic deformation method, it was possible to construct new
classes of exact cosmological solutions with generic o-diagonal metrics. We also
analyzed scenarios for Finsler acceleration of universe.
One of the most important cosmological problem which should be solved in
Finsler type gravity theories is that how we could avoid possible over-closure of the
Universe. This nontrivial issue was not studied in this paper. We suppose that real-
istic cosmological scenarios can be elaborated in a self-consistent theoretical form
and in correspondence with observational data for a certain Finsler brane cosmolog-
ical models with trapping/working as we discussed in Refs. 50, 41 and 51; at least,
we can be sure that cyclic and ekpyrotic scenarios are possible for modied Finsler
theories
58
and we have a general geometric method for generating o-diagonal cos-
mological solutions in Finsler gravity. The properties of such solutions depend on
the type of matter sources, symmetries and boundary/initial conditions we impose
for our models. For stochastic o-diagonal components of metrics,
57
with momen-
tum transfer, we reproduce the same problems for overclosure of universes as in
low energy string-brane limits with stochastic foam.
4,31
This can be considered as
a possible direction for our further research.
Finally, it is worth mentioning an orthodox approach with Finsler like and/or
almost Kahler variables when the cosmological solutions are derived for generic o-
diagonal metrics in GR.
15,41
Following this approach, we may conclude that there
are not modications of Einstein gravity at classical level and that all accelerating
and anisotropic eects in our days cosmology are consequences of certain nonlinear
o-diagonal classical gravitational and matter eld interactions. Considering (in
Finsler variables) nonholonomically deformed FRW universes, we may model the
bulk of dark energy and dark matter physics. Such exact cosmological solutions
can be constructed in explicit form.
34
Nevertheless, we have to work with canonical
Finsler gravity models on (co)tangent bundles if quantum eects are taken into con-
sideration. Further developments will be provided in our papers under elaboration.
Acknowledgment
I am grateful to N. Mavromatos and P. Stavrinos for important discussions, support
and collaboration. The research in this paper is partially supported by the Program
IDEI, PN-II-ID-PCE-2011-3-0256. Some recent results were communicated at the
parallel section Quantum Gravity Phenomenology, head G. Amelino-Camelia,
at Marcell Grossman 13, MG-13, Stockholm, Sweden (July 17, 2012). I thank also
1250072-35
S. I. Vacaru
the referee for very important critics and suggested new references which I consider
improved substantially the physical content of this paper.
Appendix A. Formulas for N-Adapted Coecients
For convenience, we present some important formulas in Finsler geometry and
applications in modern gravity (see details in Refs. 21, 26, 10 and 22). Introduc-
ing the respective h- v-components of the d-connection one-form for d-connection
D = {
}, we get the N-adapted coecients T
= {T
i
jk
, T
i
ja
, T
a
ji
, T
a
bi
, T
a
bc
}
of torsion d-tensor (12),
T
i
jk
= L
i
jk
L
i
kj
, T
i
ja
= T
i
aj
= C
i
ja
, T
a
ji
=
a
ji
,
T
a
bi
= T
a
ib
=
N
a
i
y
b
L
a
bi
, T
a
bc
= C
a
bc
C
a
cb
.
(A.1)
A N-adapted dierential form calculus allows us to derive the formulas for h-
v-components of curvature d-tensor (13) of a d-connection D, i.e. d-curvature
R
= {R
i
hjk
, R
a
bjk
, R
i
jka
, R
c
bka
, R
i
jbc
, R
a
bcd
}, when
R
i
hjk
= e
k
L
i
hj
e
j
L
i
hk
+L
m
hj
L
i
mk
L
m
hk
L
i
mj
C
i
ha
a
kj
,
R
a
bjk
= e
k
L
a
bj
e
j
L
a
bk
+L
c
bj
L
a
ck
L
c
bk
L
a
cj
C
a
bc
c
kj
,
R
i
jka
= e
a
L
i
jk
D
k
C
i
ja
+C
i
jb
T
b
ka
,
R
c
bka
= e
a
L
c
bk
D
k
C
c
ba
+C
c
bd
T
c
ka
,
R
i
jbc
= e
c
C
i
jb
e
b
C
i
jc
+C
h
jb
C
i
hc
C
h
jc
C
i
hb
,
R
a
bcd
= e
d
C
a
bc
e
c
C
a
bd
+C
e
bc
C
a
ed
C
e
bd
C
a
ec
.
(A.2)
The values (A.1) and (A.2) can be computed in explicit form for the canonical
d-connection

= (
L
i
jk
,
L
a
bk
,

C
i
jc
,

C
a
bc
) with
L
i
jk
=
1
2
g
ir
(e
k
g
jr
+e
j
g
kr
e
r
g
jk
) ,
L
a
bk
= e
b
(N
a
k
) +
1
2
h
ac
e
k
h
bc
h
dc
e
b
N
d
k
h
db
e
c
N
d
k
, (A.3)
C
i
jc
=
1
2
g
ik
e
c
g
jk
,

C
a
bc
=
1
2
h
ad
(e
c
h
bd
+e
c
h
cd
e
d
h
bc
) .
For any d-metric g on an N-anholonomic manifold V,

D = {
} satises the
condition

Dg = 0 vanishing of pure horizontal and vertical torsion coecients,
i.e.

T
i
jk
= 0 and

T
a
bc
= 0 (see formulas (A.1)). We emphasize that, in general,
T
i
ja
,

T
a
ji
and

T
a
bi
are not zero, but such nontrivial components of torsion are
induced by coecients of an o-diagonal metric g
(11).
1250072-36
Any geometric construction for the canonical d-connection

D = {
} can
be re-dened equivalently into a similar one with the LeviCivita connection =
{
} following formulas:
, (A.4)
where N-adapted coecients of connections,
and

, and the distor-

tion tensor

Z
are determined in unique forms by the coecients of a

metric g
. The N-adapted components of the distortion tensor

Z
=
{Z
a
jk
, Z
i
bk
, Z
a
bk
, Z
i
kb
, Z
i
jk
, Z
a
jb
, Z
a
bc
, Z
i
ab
} are
Z
a
jk
=
C
i
jb
g
ik
h
ab
1
2
a
jk
, Z
i
bk
=
1
2
c
jk
h
cb
g
ji
ih
jk

C
j
hb
,
Z
a
bk
=
+
ab
cd

T
c
kb
, Z
i
kb
=
1
2
a
jk
h
cb
g
ji
+
ih
jk

C
j
hb
, Z
i
jk
= 0, (A.5)
Z
a
jb
=
ad
cb

T
c
jd
, Z
a
bc
= 0, Z
i
ab
=
g
ij
2
[
T
c
ja
h
cb
+

T
c
jb
h
ca
],
for
ih
jk
=
1
2
(
i
j
h
k
g
jk
g
ih
),

ab
cd
=
1
2
(
a
c
b
d
+h
cd
h
ab
) and

T
c
ja
=

L
c
aj
e
a
(N
c
j
).
Appendix B. A Class of General Cosmological Solutions
The Einstein equations for

D computed for ansatz (25) and source
= diag[ 3
;
1
=
2
=
2
(x
k
);
3
=
4
=
4
(x
k
,
0
v);
5
=
6
=
6
(u
0
,
1
v);
7
=
8
=
8
(u
1
,
2
v)],
(B.1)
when the partial derivatives, for instance, are parametrized 1
v
= /
1
v = /y
5
,
2
v
= /
2
v = /y
7
and N
5
0
=
1
w 0
(u
0
,
1
v), N
6
0
=
1
n 0
(u
0
,
1
v),
N
7
1
=
2
w 1
(u
1
,
2
v), N
8
1
=
2
n 1
(u
1
,
2
v). For zero N-connection coecients
N
a
i
, with i, j, . . . = 1, 2, 3, 4 and a, b, . . . = 5, 6, 7, 8, we can chose such solutions for
h
a
when (44) have certain limits to the diagonal cosmological metric (37). Such
a very general o-diagonal, inhomogeneous and locally anisotropic cosmological
dynamics, with one Killing symmetry vector /y
8
= /
v
(a similar class of
solutions can be generated if as y
8
we take
v
for y
7
=
v
) is determined by
coecients
h
5
=
0
1
h(x
1
,
h
r, t,
h
) [ v
r
1
f(x
1
,
h
r, t,
h
,
v
r)]
2
|
1
(x
1
,
h
r, t,
h
,
v
r)|,
h
6
= [
1
f(x
1
,
h
r, t,
h
,
v
r)
0
1
f(x
1
,
h
r, t,
h
)]
2
,
5
= [f(
h
r, t,
h
)
0
f(
h
r, t)]
2
,

h
5
=
h
a
2
(t) (
h
r)
2
sin
2 h
,
w 0
= 0
1
(x
1
,
h
r, t,
h
,
v
r)/ v
r
1
(x
1
,
h
r, t,
h
,
v
r),
n 0
=
1
n 0
(x
1
,
h
r, t,
h
) +
2
n 0
(x
1
,
h
r, t,
h
)
d
v
r
1
(x
1
,
h
r, t,
h
,
v
r)
[ v
r
1
f(x
1
,
h
r, t,
h
,
v
r]
2
[
1
f(x
1
,
h
r, t,
h
,
v
r)
0
1
f(x
1
,
h
r, t,
h
)]
3
,
1250072-37
S. I. Vacaru
for
1
=
0
1
(x
1
,
h
r, t,
h
)
1
2
8
0
1
h(x
1
,
h
r, t,
h
)
d
v
r [ 1
v
1
f(x
1
,
h
r, t,
h
,
v
r)]
[
1
f(x
1
,
h
r, t,
h
,
v
r)
0
1
f(x
1
,
h
r, t,
h
)]; for any generation
1
f(x
1
,
h
r, t,
h
,
v
r)]
2
and integration functions
0
1
h(x
1
,
h
r, t,
h
) and small parameter ,
0
1
h(x
1
,
h
r, t,
h
)
[ 1
v
1
f(x
1
,
h
r, t,
h
,
v
r)]
2
|
1
(x
1
,
h
r, t,
h
,
v
r)| =
v
a
2
(t)(1+(x
1
,
h
r, t,
h
,
v
r)),
and w 1
= 0, n 1
= 0, h
7
=
v
a
2
(t)(
v
r)
2
, h
8
=
v
a
2
(t) (
v
r)
2
sin
2 v
.
bb
For simplicity, we can x
1
n 0
= 0 and that
1
f(x
1
,
h
r, t,
h
,
v
r) induces
= (t,
h
,
v
r) as a solution of any three dimensional solitonic (nonlinear wave)
equation, for instance, of type

2
(
v
r)
2
+ (
+ 6
= 0, = 1, where
= /t and
= /
h
. Such solitons are stable and generate solitonic
congurations for the metric and nontrivial N-connection coecients (for
0
= 3, 4),

5
=
5
[] 1+
5
[],
6
=
6
[] 1+
6
[] and w
3
w
3
[], w
4
w
4
[], n
3

n
3
[], n
4
n
4
[], where, for simplicity, we xed the boundary conditions to have
a functional dependence on and vanishing values if 0.
References
1. J. D. Bekenstein, Phys. Rev. D 48 (1993) 3641.
2. C. Will, Theory and Experiment in Gravitational Physics, Revised edn. (Cambridge
University Press, Cambridge, 1993).
3. N. E. Mavromatos, Phys. Rev. D 83 (2011) 025018.
4. N. E. Mavromatos, Phys. Lett. B 696 (2011) 300.
5. C. Lammerzahl, D. Lorek and H. Dittus, Gen. Relativ. Gravit. 41 (2009) 1345.
6. S. Vacaru, Phys. Lett. B 690 (2010) 224.
7. F. Girelli, S. Liberati and L. Sindoni, Phys. Rev. D 75 (2007) 064015.
8. L. Sindoni, Phys. Rev. D 77 (2008) 124009.
9. J. Skakala and M. Visser, J. Phys. Conf. Ser. 189 (2009) 012037.
10. S. Vacaru, Int. J. Geom. Methods Mod. Phys. 5 (2008) 473.
11. A. P. Kouretsis, M. Stathakopoulos and P. C. Stavrinos, Phys. Rev. D 79 (2009)
104011.
12. P. C. Stavrinos, A. P. Kouretsis and M. Stathakopoulos, Gen. Relativ. Gravit. 40
(2008) 1403.
13. Z. Chang and X. Li, Phys. Lett. B 663 (2008) 103.
14. Z. Chang and X. Li, Phys. Lett. B 656 (2009) 173.
16. E. Cartan, Les Espaces de Finsler (Paris, Herman, 1935).
17. J. I. Horvath, Phys. Rev. 80 (1950) 901.
18. M. Matsumoto, Foundations of Finsler Geometry and Special Finsler Spaces (Shin-
gaken, Kaisisha, Japan, 1986).
19. R. Miron and M. Anastasiei, The Geometry of Lagrange Spaces: Theory and Applica-
tions (Kluwer Academic Publishers, Dordrecht, Boston, London, 1994).
20. S. Vacaru, Nucl. Phys. B 434 (1997) 590.
21. S. Vacaru, J. Math. Phys. 46 (2005) 042503.
bb
We considered, for example, only small contributions from the velocity h-subspace to the real
v-subspaces.
1250072-38
22. S. Vacaru, P. Stavrinos, E. Gaburov and D. Gont a, Dierential Geometry Dynam-
ical Systems (Geometry Balkan Press, 2006), www.mathem.pub.ro/dgds/mono/va-
t.pdf & arXiv:gr-qc/0508023.
23. D. Bao, S.-S. Chern and Z. Shen, An Introduction to RiemannFinsler Geometry,
Graduate Texts in Math., Vol. 200 (Springer-Verlag, 2000).
24. P. Finsler,

Uber Kurven und Fl achen in allgemeinen Raumen, Dissertation, University
of Gottingen, printed 1951 (Birkh auser, Basel, 1918).
25. B. Riemann, Habilitationsvorlesung, University of Gottingen, published 1892 (Leipzig,
1854).
27. G. Amelino-Camelia, Int. J. Mod. Phys. D 11 (2002) 35.
28. J. Maueijo and L. Smolin, Class. Quantum Grav. 21 (2004) 1725.
29. D. Kimberly, J. Magueijo and J. Medeiros, Phys. Rev. D 70 (2004) 084007.
30. J. Ellis, N. E. Mavromatos and D. V. Nanopoulos, Gen. Relativ. Gravit. 32 (2000)
127.
31. N. E. Mavromatos, V. A. Mitsou, S. Sarkar and A. Vergou, Euro. Phys. J. C 72
(2012) 1956.
32. G. W. Gibbons, J. Gomis and C. N. Pope, Phys. Rev. D 76 (2007) 081701.
33. S. Vacaru, Int. J. Theor. Phys. 49 (2010) 884.
34. S. Vacaru, Int. J. Theor. Phys. 49 (2010) 2753.
35. V. Perlick, Gen. Relativ. Gravit. 38 (2006) 365.
36. S. Mignemi, Phys. Rev. D 76 (2007) 047702.
37. C. W. Misner, K. S. Thorne and J. A. Wheeler, Gravitation (Freeman, 1973).
38. L. D. Landau and E. M. Lifshitz, The Classical Theory of Fields, Vol. 2, 2nd and 3rd
edn. (Pergamon, London, 1962 & 1967) ([4th edition does not contain the imaginary
unity for pseudo-Euclidean metrics]).
39. C. Mller, Theory of Relativity, 2d edn. (Oxford University Press, 1972).
40. A. Bejancu and H. R. Farran, Geometry of Pseudo-Finsler Submanifolds (Kluwer
Academic Publishers, 2000).
41. S. Vacaru, Class. Quantum Grav. 27 (2010) 105003.
42. S. Vacaru, Phys. Lett. A 372 (2008) 2949.
44. D. Colladay and V. A. Kostelecky, Phys. Rev. D 55 (1997) 6760.
45. J. Ehlers, F. A. Pirani and A. Schild, The geometry of free fall and light propagation,
in General Relativity. Papers Honour of J. L. Synge, ed. L. ORaifertaigh (Clarendon
Press, Oxford, 1972), p. 63.
46. F. A. E. Pirani, Building spacetime from light rays and free particles, in Symposia
Mathematica, Vol. XII (Academic Press, NY, 1973), p. 67.
47. N. M. J. Woodhouse, J. Math. Phys. 14 (1973) 495.
48. R. Meister, Eine neuformulierung der EPS axiomatik, masters thesis (Paderborn
University, Paderborn, 1990).
49. R. I. Pimenov, Anisotropic Finsler Extension of General Relativity as an Order Struc-
ture (Syktyvkar, Komi Filial, 1987) (Russian).
50. S. Vacaru, Class. Quantum Grav. 28 (2011) 215991.
51. S. Vacaru, Gen. Relativ. Gravit. 44 (2012) 1015.
52. S. Vacaru and V. Stavrinos, Spinors and Space-Time Anisotropy (Athens University
Press, Athens, Greece, 2002).
53. M. Anastasiei and S. Vacaru, J. Math. Phys. 50 (2009) 013510.
54. S. Ikeda and M. Anastasiei, Balkan J. Geom. Appl. 11 (2006) 81.
1250072-39
S. I. Vacaru
55. A. P. Kouretsis, M. Stathakopoulos and P. C. Stavrinos, Phys. Rev. D 82 (2010)
064035.
56. S. Vacaru and D. Singleton, Class. Quantum Grav. 19 (2002) 3583.
57. S. Vacaru, Eur. Phys. J. 127 (2012) 32, arXiv:1010.0647 [math-ph].
58. P. C. Stavrinos and S. Vacaru, Cyclic and ekpyrotic universes in modied Finsler
osculating gravity on tangent Lorentz bundles, arXiv:1206.3998.
1250072-40

Principles of Einstein-Finsler Gravity and Perspectives in Modern Cosmology

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Principles of Einstein-Finsler Gravity and Perspectives in Modern Cosmology

Uploaded by

Copyright:

Available Formats

October 2, 2012 3:46 WSPC/S0218-2718 142-IJMPD 1250072

International Journal of Modern Physics D

=0. We can write for some classes of co-ordinate

with (antisymmetric) nontrivial anholonomy coecients w

, where i is the imaginary

). Euclidean co-ordinates with

(u), we can transform

} on V in a standard form. Nevertheless, this connection is not used in Finsler

and parametrized in the

, when the coecients of

} is dened in a standard form by contracting

, The h-/v-components of this

} and the distorting

} are uniquely dened by the same metric tensor g. The coef-

(14) and Einstein E

(15) d-tensors. The N-adapted grav-

, where the source

are derived in such a way that they contain contributions

(matter energymomentum in Einstein gravity) if

(A.5) are zero. For such congurations, we have

} satisfying the eld equations

]. Such equations can be solved

, we can dene such N-adapted frames when the coecients of

D are equal to the coecients of the Cartan d-connection

has well dened

= 0, and integration functions

Dg = 0 and h- and v-torsions vanish but the general nonzero torsion

Z (16). So, we can describe geometrical and physical models equiva-

V, are well-dened. For corresponding

V. We can consider such theories as a simple framework

on TV for the gen-

2. It is always satised the condi-

> 0, for <

b, . . . = 4, 5 (such a model describes evolution in time t of a h-subspace for

= /. The conditions for

. Conclusions about Finsler-solitonic o-

() and (). Respectively, the solitonic versions of Eq. (39) are

are, in general, dierent. Small modications

may transfer, for instance, an accelerating conguration

}, we get the N-adapted coecients T

, and the distor-

are determined in unique forms by the coecients of a

. The N-adapted components of the distortion tensor

You might also like