GAUGE GRAVITATION THEORY.

WHAT IS GEOMETRY OF THE WORLD?

Gennadi A Sardanashvily
Department of Theoretical Physics, Physics Faculty,
Moscow State University, 117234 Moscow, Russia
E-mail: [email protected]
When joined the unied gauge picture of fundamental interactions, the gravitation
theory leads to geometry of a space-time which is far from simplicity of pseudo-Riemannian
geometry of Einstein's General Relativity. This is geometry of the ane-metric composite
dislocated manifolds. The goal is modication of the familiar equations of a gravitational
eld and entirely the new equations of its deviations. In the present brief, we do not detail
the mathematics, but discuss the reasons why it is just this geometry. The major physical
underlying reason lies in spontaneous symmetry breaking when the fermion matter admits
only the Lorentz subgroup of world symmetries of the geometric arena.
I.
Gauge theory is well-known to call into play the dierential geometric methods in
order to describe interaction of elds possessing a certain symmetry group. Moreover,
it is this geometric approach phrased in terms of bred manifolds which provides the
adequate mathematical formulation of classical eld theory.
Note that the conventional gauge principle of gauge invariance of Lagrangian densities
losts its validity since symmetries of realistic eld models are almost always broken. We
follow the geometric modication of the gauge principle which is formulated as follows.
Classical elds can be identied with sections of bred manifolds
:Y !X (1)
whose n-dimensional base X is treated a parameter space, in particular, a world manifold.
A locally trivial bred manifold is called the bre bundle (or simply the bundle). We
further provide Y with an atlas of bred coordinates (x; yi) where (x) are coordinates
of its base X .
GR-QC-9410045

Classical eld theory meets the following tree types of elds as a rule:
 matter elds represented by sections of a vector bundle
Y = (P  V )=G
associated with a certain principal bundle P ! X whose structure group is G;

1
  gauge potentials identied to sections of the bundle
C := J 1P=G ! X (2)
of principal connections on P (where J 1P denotes the rst order jet manifold of the
bundle P ! X );
 classical Higgs elds described by global sections of the quotient bundle
K := P=K ! X
where K is the exact symmetry closed subgroup of the Lie group G.
Dynamics of elds represented by sections of the bred manifold (1) is phrased in
terms of jet manifolds [2, 3, 9, 10, 20, 22].
Recall that the k-order jet manifold J k Y of a bred manifold Y ! X comprises the
equivalence classes jxk s, x 2 X , of sections s of Y identied by the (k + 1) terms of their
Taylor series at x. It is important that J k Y is a nite-dimensional manifold which meets
all conditions usually required of manifolds in eld theory. Jet manifolds have been widely
used in the theory of dierential operators. Their application to dierential geometry is
based on the 1:1 correspondence between the connections on a bred manifold Y ! X
and the global sections
= dx
(@ + i(y)@i) (3)
of the jet bundle J 1Y ! Y [11, 20, 27]. The jet bundle J 1Y ! Y is an ane bundle
modelled on the vector bundle
J 1Y = T X
Y V Y
where V Y denotes the vertical tangent bundle of the bred manifold Y ! X . It follows
that connections on a bred manifold Y constitute an ane space modelled on the vector
space of soldering forms Y ! J 1Y on Y .
In the rst order Lagrangian formalism, the jet manifold J 1Y plays the role of a nite-
dimensional conguration space of elds. Given bred coordinates (x; yi) of Y ! X , it
is endowed with the adapted coordinates (x; yi; yi ) where coordinates yi make the sense
of values of rst order partial derivatives @yi(x) of eld functions yi(x). In jet terms, a
rst order Lagrangian density of elds is represented by an exterior horizontal density
L = L(x; yi; yi )!; ! = dx1 ^ ::: ^ dxn; (4)
on J 1Y ! X .
In eld theory, all Lagrangian densities are polynomial forms with respect to velocities
yi . Since the jet bundle J 1Y ! Y is ane, polynomial forms on J 1Y are factorized by
morphisms J 1Y ! J 1Y . It follows that every Lagrangian density of eld theory is
represented by composition
L : J 1Y ! T X
Y V Y ! ^n T X
D 
(5)

2
where D is the covariant dierential
D = (yi i)dx
@i
with respect to some connection (3) on Y ! X . It is the fact why the gauge principle and
the variation principle involve connections on bred manifolds in order to describe eld
systems. In case of the standard gauge theory, they are principal connections treated the
mediators of interaction of elds.
Note that several equivalent denitions of connections on bred manifolds are utilized.
We follow the general notion of connections as sections of jet bundles, without appealing to
transformation groups. This general approach is suitable to formulate the classical concept
of principal connections as sections of the jet bundle J 1P ! P which are equivariant
under the canonical action of the structure group G of P on P on the right. Then, we
get description of principal connections as sections of the bundle (2).
II.
In comparison with case of internal symmetries, the gauge gravitation theory meets
two objects. These are the fermion matter and the geometric arena. They might arise
owing to sui generis primary phase transition which had separated prematter and prege-
ometry. One can think of the well-known possibility of describing a space-time in spinor
coordinates as being the relic of that phase transition.
Here, we are not concerned with these quetions, and by a world manifold is meant
a 4-dimensional oriented manifold X 4 coordinatized in the standard manner. By world
geometry is called dierential geometry of the tangent bundle TX and the cotangent
bundle T X of X 4. The structure group of these bundles is the general linear group
GL4 = GL(4; R):
The associated principal bundle is the bundle LX ! X of linear frames in tangent spaces
to X 4.
Note that, on the physical level, a basis of the tangent space Tx to X 4 at x 2 X 4 is
usually interpreted as the local reference frame at a point. However, realization of such a
reference frame by physical devices remains under discussion. A family fzg of local sec-
tions of the principal bundle LX sets up an atlas of LX (and so the associated atlases of
TX and T X ) which we, following the gauge theory tradition, treat sui generis the world
reference frame. In gauge gravitation theory in comparison with case of internal sym-
metries, there exists the special subclass of holonomic atlases f T g whose trivialization
morphisms are the tangent morphisms
 = T
T

to trivialization morphisms  of coordinate atlases of X 4. The associated bases of TX

and T X are the holonomic bases f@g and fdxg respectively. Note that Einstein's Gen-
eral Relativity was formulated just with respect to holonomic reference frames, so that
3
many mixed up reference frames and coordinate systems. The mathematical reason of
separating holonomic reference frames consists in the fact that the jet manifold formal-
ism is phrased only in terms of holonomic atlases. However, holonomic atlases are not
compatible generally with spinor bundles whose sections describe the fermion elds.
Dierent spinor models of the fermion matter have been suggested. But all observable
fermion particles are the Dirac fermions. There are several ways in order to introduce
Dirac fermion elds. We do this as follows.
Given a Minkowski space M with the Minkowski metric , let C 1;3 be the complex
Cliord algebra generated by elements of M . A spinor space V is dened to be a minimal
left ideal of C 1;3 on which this algebra acts on the left. We have the representation
:M
V !V (6)
of elements of the Minkowski space M by Dirac's matrices on V . Let us consider
the transformations preserving the representation (6). These are pairs (l; ls) of Lorentz
transformations l of the Minkowski space M and invertible elements ls of C 1;3 such that
(lM
lsV ) = ls (M
V ):
Elements ls form the Cliord group whose action on M however is not eective, therefore
we restrict ourselves to its spinor subgroup Ls = SL(2; C ).
Let us consider a bundle of complex Cliord algebras C 3;1 over X 4. Its subbundles
are both a spinor bundle SM ! X 4 and the bundle YM ! X 4 of Minkowski spaces of
generating elements of C 3;1 . To describe Dirac fermion elds on a world manifold, one
must require that YM is isomorphic to the cotangent bundle T X of a world manifold X 4.
It takes place if the structure group of LX is reducible to the Lorentz group L = SO(3; 1)
and LX contains a reduced L subbundle Lh X such that
YM = (Lh X  M )=L:
In this case, the spinor bundle SM is associated with the Ls -lift Ph of Lh X :
SM = Sh = (Ph  V )=Ls : (7)
In accordance with the well-known theorem, there is the 1:1 correspondence between
the reduced subbubdles Lh X of LX and the tetrad gravitational elds h identied with
global sections of the quotient bundle
 := LX=L ! X 4:
This bundle is the 2-fold cover of the bundle of pseudo-Riemannian bilinear forms in
cotangent spaces to X 4. Global sections of the latter are pseudo-Riemannian metrics g
on X 4. Thus, existence of a gravitational eld is necessary condition in order that Dirac
fermion elds live on a world manifold.
4
 Note that a world manifold X 4 must satisfy the well-known global topological con-
ditions in order that gravitational elds, space-time structure and spinor structure can
exist. To summarize these conditions, one may assume that X 4 is not compact and the
linear frame bundle LX over X 4 is trivial.
Given a tetrad eld h, let h be an atlas of LX such that the corresponding local
sections zh of LX take their values into the reduced subbundle LhX . This atlas has
L-valued transition functions. It fails to be holonomic in general. Moreover, relative to
h, the pseudo-Riemannian metric g corresponding to h comes to the Minkowski metric
and h takes its values in the center 0 of the quotient space GL4=L. With respect to an
atlas h and a holonomic atlas T = f T g of LX , the tetrad eld h can be represented
by a family of GL4-valued tetrad functions
h = T  zh;
dx = ha(x)ha;
which carry atlas (gauge) transformations between bre bases fdxg and fhag of T X
associated with T and h respectively. The well-known relation
g = ha hb ab (8)
takes place.
It is the feature only of a tetrad gravitational eld that it itself determines reference
frames.
Given a tetrad eld h, one can dene the representation
h : T X
Sh = (Ph  (M
V ))=Ls ! (Ph  (M
V ))=Ls = Sh ; (9)
dx
b  = (dx ) = h(x) a ;
h a
of cotangent vectors to a world manifold X by Dirac's -matrices on elements of the
4
spinor bundle Sh.
Let Ah be a connection on Sh associatede with a principal connection on LhX and D
the corresponding covariant dierential. Given the representation (9), one can construct
the Dirac operator
Dh = h  D : J 1Sh ! T X S
h V Sh ! V Sh
on Sh . Then, we can say that sections of the spinor bundle Sh describe Dirac fermion
elds in the presence of the tetrad gravitational eld h and the gauge potential Ah.
Thus, the geometry of the gauge gravitation theory is the metric-ane geometry
characterized by the pair (h; Ah) of a tetrad eld h and a reduced Lorentz connection Ah
treated as the gauge gravitational potential. Note that the metric-ane geometry is an
attribute of all gauge approaches to gravitation theory [6, 12]. Moreover, it is also the
Klein-Chern geometry of Lorentz invariants.
5
 III.
If to forget fermion elds for a moment, one can derive the gauge gravitation theory
from the equivalence principle reformulated in geometric terms [7, 18].
In Einstein's General Relativity, the equivalence principle is called to provide transition
to Special Relativity with respect to some reference frames. In the spirit of F.Klein's
Erlanger program, the Minkowski space geometry can be characterized as geometry of
Lorentz invariants. The geometric equivalence principle then postulates that there exist
reference frames with respect to wich Lorentz invariants can be dened everywhere on a
world manifold X 4. This principle has the adequate mathematical formulation in terms
of bre bundles. It requires that the linear frame bundle LX is reducible to a Lorentz
subbundle LhX whose structure group is L. They are atlases of Lh X with respect to which
Lorentz invariants can be dened, and the pseudo-Riemannian metric g corresponding to
h exemplies such a Lorentz invariant.
Thereby, the geometric equivalence principle provides a world manifold with the so-
called L-structure [26]. From the physical point of view, it singles out the Lorentz group
as the exact symmetry subgroup of world symmetries broken spontaneously [7]. The
associated classical Higgs eld is a tetrad gravitational eld.
Note that the geometric equivalence principle sets au also a space-time structure on a
world manifold X 4. In virtue of the well-known theorems, if the structure group of LX is
reducible to the structure Lorentz group, the latter, in turn, is reducible to its maximal
compact subgroup SO(3). Moreover, we have the commutative diagram
GL4 ! SO(4)
? ? (10)
L ! SO(3)
of structure groups of LX . It follows that, for every reduced subbundle LhX , there exist
a reduced subbundle FX of LX with the structure group SO(3) and the corresponding
(3+1) space-time decomposition
TX = FX  T 0X
of the tangent bundle of X 4 into a 3-dimensional spatial distribution FX and its time-like
orthocomplement T 0X .
In other words, we have also the Klein-Chern geometry of spatial invariants on a world
manifold. One can always choose an atlas of LX whose transition functions are SO(3)-
valued, and spatial invariants are exemplied by the global eld of local time directions
T 0X .
Recall that there is the 1:1 correspondence
FX c
= 0
between the nonvanishing 1-forms
on a manifold X and the 1-codimensional distribu-
tions on X .
Then, we get the following modication of the well-known theorem [18, 23].
6
  For every gravitational eld g on a world manifold X 4, there exists an associated
pair (FX; gR) of a space-time distribution FX generated by a tetrad 1-form
h0 = h0dx
and a Riemannian metric gR, so that
gR = 2h0
h0 g: (11)
Conversely, given a Riemannian metric gR, every oriented smooth 3-dimensional
distribution FX with a generating form
is a space-time distribution compatible
with the gravitational eld g given by expression (11) where
h0 = j

j ; j
j2 = gR(
;
) = g(
;
):

The triple (g; FX; gR) (11) sets up uniquely a space-time structure on a world manifold.
In particular, if the generating form of a space-time distribution FX is exact, we have
the causal space-time foliation of X 4 what corresponds exactly to the stable causality by
Hawking.
A Riemannian metric gR in the triple (11) denes a g-compatible distance function on a
world manifold X 4. Such a function turns X 4 into a metric space whose locally Euclidean
topology is equivalent to the manifold topology on X 4. Given a gravitational eld g, the g-
compatible Riemannian metrics and the corresponding distance functions are dierent for
dierent space-time distributions FX . It follows that physical observers associated with
dierent distributions perceive the same world manifold as dierent Riemannian spaces.
The well-known relativistic changes of sizes of moving bodies exemplify this phenomenon.
One often loses sight of the fact that a certain Riemannian metric and, consequently,
a metric topology can be associated with a gravitational eld [18]. For instance, one
attempts to derive a world topology directly from pseudo-Riemannian structure of a
space-time. These are path topology etc. [5]. If a space-time obeys the strong causality
condition, such topologies coincide with the familiar manifold topology of X . In general
case, they however are rather extraordinary.
IV.
A glance on the diagramm (10) shows that, in the gravitation theory, we have a
collection of spontaneous symmetry breakings:
 GL4 ! L where the corresponding Higgs eld is a gravitational eld;
 L ! SO(3) where the Higgs-like eld is represented by the tetrad 1-form h0 as a
global section of the quotient bundle
Lh X=SO(3) ! X 4;
7
  GL4 ! SO(4) where the Higgs-like eld is the Riemannian metric >from expression
(11).
Spontaneous symmetry breaking is quantum phenomenon modelled by a Higgs eld. In
the algebraic quantum eld theory, Higgs elds characterize nonequivalent Gaussian states
of algebras of quantum elds. They are sui generis ctitious elds describing collective
phenomena. In the gravitation theory, spontaneous symmetry breaking displays on the
classical level and the feature of a gravitational eld is that it is a dynamic Higgs eld.
Indeed, the splitting (8) of the metric eld looks like the standard decomposition of a
Higgs eld where the Minkowski metric  and the tetrad functions play the role of the L-
stable vacuum Higgs eld and the Goldstone elds respectively. However, in contrast with
the internal symmetry case, the Goldstone components of a gravitational eld can not be
removed by gauge transformations because the reference frames h fail to be holonomic
in general and, roughly speaking, the associated basis elements ha = ha@ contain tetrad
functions.
>For the rst time, the conception of a graviton as a Goldstone particle corresponding
to violation of Lorentz symmetries in a curved space-time had been advanced in mid 60s
by Heisenberg and Ivanenko in discussion on cosmological and vacuum asymmetries. This
idea was revived in connection with constructing the induced representations of the group
GL4 and then in the framework of the approach to gravitation theory as a nonlinear -
model [6, 13, 14]. In geometric terms, the fact that a pseudo-Riemannian metric is similar
to a Higgs eld has been pointed out by A.Trautman and by us [7].
The Higgs character of classical gravity is founded on the fact that, for dierent tetrad
elds h and h0, Dirac fermion elds are described by sections of spinor bundles associated
with dierent reduced L-principal subbundles of LX and so, the representations h and
h (9) are not equivalent [18]. It follows that e Dirac fermion eld must be regarded only
0

in a pair with a certain tetrad gravitational eld h. These pairs constitute the so-called
fermion-gravitation complex [13]. They can not be represented by sections of any product
S   where S ! X 4 is some standard spinor bundle. At the same time, there is the 1:1
correspondence between these pairs and the sections of the composite spinor bundle
S !  ! X4 (12)
where S !  is a spinor bundle associated with the L principal bundle LX !  [18, 24].
In particular, every spinor bundle Sh (7) is isomorphic to restriction of S to h(X 4)  .
V.
By a composite manifold is meant the composition
Y !!X (13)
where Y !  is a bundle denoted by Y and  ! X is a bred manifold.

8
 In analytical mechanics, composite manifolds
Y !!R
characterize systems with variable parameters, e.g. the classical Berry's oscillator [19]. In
gauge theory, composite manifolds
P ! K ! X
describe spontaneous symmetry breaking [17, 20].
Application of composite manifolds to eld theory is founded on the following spec-
ulations. Given a global section h of , the restriction Yh of Y to h(X ) is a bred
submanifold of Y ! X . There is the 1:1 correspondence between the global sections sh of
Yh and the global sections of the composite manifold (13) which cover h. Therefore, one
can say that sections sh of Yh describe elds in the presence of a background parameter
eld h, whereas sections of the composite manifold Y describe all pairs (sh; h). It is im-
portant when the bundles Yh and Yh6=h fail to be equivalent in a sense. The conguration
0

space of these pairs is the rst order jet manifold J 1Y of the composite manifold Y .
The feature of the dynamics of eld systems on composite manifolds consists in the
following [20, 25].
Let Y be a composite manifold (13) provided with the bred coordinates (x; m; yi)
where (x; m) are bred coordinates of . Every connection
A = dx
(@ + Aei@i) + dm
(@m + Aim@i)
on Y !  yields splitting
V Y = V Y Y (Y  V )
and, as a consequence, the rst order dierential operator
f : J 1Y ! T  X
V Y ;
D Y
D = dx
(y Aei Aimm)@i;
f  i

on Y . Let h be a global section of  and Yh the restriction of the bundle Y to h(X ).

f to J 1 Y  J 1Y comes to the familiar covariant dierential relative
The restriction of D h
to a certain connection Ah on Yh . Thus, it is D
f that one may utilize in order to construct
a Lagrangian density (5)
T X
Y V Y ! ^n T X
De 
L : J 1Y !
for sections of a composite manifold. It should be noted that such a Lagrangian density
is never regular because of the constraint conditions
Aim@iL = @m L:
9
 Recall that, if a Lagrangian density is degenerate, the corresponding Euler-Lagrange
equations are underdetermined and need supplementary gauge-type conditions which
remain elusive in general. Therefore, to describe constraint eld systems, we utilize
the multimomentum Hamiltonian formalism where canonical momenta correspond to
derivatives of elds with respect to all world coordinates, not only the temporal one
[1, 4, 19, 20, 21, 22]. Note that application of the conventional Hamiltonian formalism
to eld theory fails to be successful. In the straightforward manner, it leads to innite-
dimensional phase spaces. In the framework of the multimomentum approach, the phase
space of elds is the Legendre bundle
n
 = ^ T X
Y TX
Y V Y (14)
over Y . It is provided with the bred coordinates (x; yi; pi). Note that every Lagrangian
density L on J 1Y determines the Legendre morphism
Lb : J 1Y ! ;
(x; yi; pi)  Lb = (x; yi; @iL):
The Legendre bundle (14) carries the multisymplectic form

= dpi ^ dyi ^ !
@:
We say that a connection on the bred Legendre manifold  ! X is a Hamiltonian
connection if the form c
is closed. Then, a Hamiltonian form H on  is dened to be
an exterior form such that
dH = c
(15)
for some Hamiltonian connection . The key point consists in the fact that every Hamil-
tonian form admits splitting
H = pidyi ^ ! pi i! Hf ! = pidyi ^ ! H!; ! = @c!; (16)
where is a connection on the bred manifold Y and Hf ! is a horizontal density on
 ! X . Given the Hamiltonian form (16), the equality (15) comes to the Hamilton
equations
@ri = @i H;
@ri = @iH (17)
for sections r of the bred Legendre manifold  ! X .
We thus observe that the multimomentum Hamiltonian formalism exemplies the
generalized Hamiltonian dynamics which is not merely a time evolution directed by the
Poisson bracket, but it is governed by partial dierential equations (17) where temporal
and spatial coordinates enter on equal footing. Maintaining covariance has the principal
10
advantages of describing eld theories, for any preliminary space-time splitting shades the
covariant picture of eld constraints. Contemporary eld models are almost always the
constraint ones.
We shall say that the Hamiltonian form H (16) is associated with a Lagrangian density
L (4) if H satises the relations which take the coordinate form
H(yj ; pj ) = pi@i H L(yj ; @j H):
If a Lagrangian density L is regular, there exists the unique Hamiltonian form H such
that the rst order Euler-Lagrange equations and the Hamilton equations are equivalent,
otherwise in general case. One must consider a family of dierent Hamiltonian forms H
associated with the same degenerate Lagrangian density L in order to exaust solutions
of the Euler-Lagrange equations. Lagrangian densities of eld models are almost always
quadratic and ane in derivative coordinates yi . In this case, given an associated Ha-
miltonian form H , every solution of the corresponding Hamilton equations which lives
on Lb (J 1Y )   yields a solution of the Euler-Lagrange equations. Conversely, for any
solution of the Euler-Lagrange equations, there exists the corresponding solution of the
Hamilton equations for some associated Hamiltonian form. Obviously, it lives on Lb (J 1Y )
which makes the sense of the Lagrangian constraint space.
The feature of Hamiltonian systems on composite manifolds (13) lies in the fact that
the Lagrangian constraint space is
pm + Aimpi = 0 (18)
[20, 21, 25]. Moreover, if h is a global section of  ! X , the submanifold h of  given
by the coordinate relations
m = hm(x); pm + Aimpi = 0
is isomorphic to the Legendre bundle over the restriction Yh of Y to h(X ). The Legendre
bundle h is the phase space of elds in the presence of the background parameter eld
h.
VI.
In gravitation theory, we have the composite manifold
LX !  ! X 4
and the associated composite spinor bundle (12). Roughly speaking, values of tetrad
gravitational elds play the role of coordinate parameters, besides the familiar world
coordinates.
In the multimomentum Hamiltonian gravitation theory, the constraint condition (18)
takes the form
pc + 81 cba(yB [ a; b]AB pA + pA
+ [ a ; b ] A yB ) = 0
+B + (19)

11
where (c; yA) are tetrad and spinor coordinates of the composite spinor bundle (12),
and pc
 and pA are the corresponding momenta [24, 25]. The condition (19) replaces the

standard gravitational constraints
pc = 0: (20)
The crucial point is that, when restricted to the constraint space (20), the Hamilton
equations (17) of gravitation theory come to the familiar gravitational equations
Ga + Ta = 0
where T denotes the energy-momentum tensor of fermion elds, otherwise on the modied
constraint space (19). In the latter case, we have the modied gravitational equations of
the total system of fermion elds and gravity:
Dpa
 = G + T
a a

where D denotes the covariant derivative with respect to the Levi-Civita connection
which acts on the indices a.
VII.
Since, for dierent tetrad elds h and h0, the representations h and h (9) are not
0

equivalent, even weak gravitational elds, unlike matter elds and gauge potentials, fail
to form an ane space modelled on a linear space of deviations of some background eld.
They thereby do not satisfy the superposition principle and can not be quantized by
usual methods, for in accordance with the algebraic quantum eld theory quantized elds
must constitute a linear space. This is the common feature of Higgs elds. In algebraic
quantum eld theory, dierent Higgs elds correspond to nonequivalent Gaussian states
of a quantum eld algebra. Quantized deviations of a Higgs eld can not change a state
of this algebra and so, they fail to generate a new Higgs eld.
At the same time, one can examine superposable deviations  of a tetrad gravitational
eld h such that h +  is not a tetrad gravitational eld [16, 18]. In the coordinate form,
such deviations read
he a = H bahb = (ab + ba)hb = H  ha = ( +  )ha = ha + a; (21)
h = g  hb = H h;
e a ab e   a
he ahe a 6= ; he ahe b 6= ab :
Note that the similar factors have been investigated by R.Percacci [15].
In bundle terms, we can describe the deviations (21) as the special morphism 1
of the cotangent bundle [16, 18]. Given a gravitational eld h and the corresponding
representation morphism h (9), the morphism 1 yields another -matrix representation
eh = h  1;
eh (ha) = H a b h(hb ) = H ab b ;
12
of cotangent vectors, but on the same spinor bundle Sh. Therefore, deviations (21) and
their superposition  + 0 can be dened.
Let us note that, to construct a Lagrangian density of deviations  of a gravitational
eld, one usually utilize a familiar Lagrangian density of a gravitationsl eld h0 = h + 
where h is treated as a background eld. In case of the deviations (21), one can not follow
this method, for quantities he fail to be true tetrad elds. To overcome this diculty,
we use the fact that the morphisms 1 appears also in the dislocation gauge theory of
the translation group. We therefore may apply the Lagrangian densities of this theory in
order to describe deviations  (21).
Let the tangent bundle TX be provided with the canonical structure of the ane
tangent bundle. It is coordinatized by (x; u) where u 6= x_  are the ane coordinates.
Every ane connection A on TX is brought into the sum
A= + (22)
of a linear connection and a soldering form
 =  (x)@
dx
which plays the role of a gauge translation potential.
In the conventional gauge theory of the ane group, one faces the problem of physical
interpretation of both gauge translation potentials and sections u(x) of the ane tangent
bundle TX . In eld theory, no elds possess the transformation law
u(x) ! u(x) + a
under the Poincare translations.
At the same time, one observes such elds in the gauge theory of dislocations [8] which
is based on the fact that, in the presence of dislocations, displacement vectors uk ; k =
1; 2; 3; of small deformations are determined only with accuracy to gauge translations
uk ! uk + ak (x):
In this theory, gauge translation potentials k i describe the plastic distortion, the covariant
derivatives
Di uk = @iuk k i
consist with the elastic distortion, and the strength
F k ij = @ik j @j k i
the dislocation density. Equations of the dislocation theory are derived from the gauge
invariant Lagrangian density
L = Di uk Di uk + 2 Di uiDm um F kij Fk ij (23)

13
where  and  are the Lame coecients of isotropic media. These equations however are
not independent of each other since a displacement eld uk (x) can be removed by gauge
translations and, thereby, it fails to be a dynamic variable.
In the spirit of the gauge dislocation theory, we have suggested that gauge potentials of
the Poincare translations may describe new geometric structure (sui generis dislocations)
of a world manifold [16, 18].
Let the tangent bundle TX be provided with an ane connection (22). By dislocation
morphism of a world manifold X is meant the special bundle isomorphism of TX over X
which takes the coordinate form
 :   @x@  ! @x@  + ( 
 u
 + 
)
@ !  ( +  ) @ =  H  @ ;
@u  
@x  
@x (24)
where
 = D uju=0 = (@u + 
 u +   )ju=0
  (25)
is the covariant derivatives of a displacement eld u.
Let Y be a bundle over X and J 1Y the jet manifold of Y . The deformation mor-
phism (24) has the jet prolongation
j 1 : yi ! H (x)yi
over Y . Then, given a Lagrangian density L of elds on a world manifold, one can think
of the composition
Le = L  
as being the corresponding Lagrangiam density of elds on the dislocated manifold. If the
above-mentioned morphism 1 consists with the dual to the morphism (24), one can apply
these Lagrangian densities in order to describe elds in the presence of the deviations (21).
Moreover, we may assume that the deviations  (21) have the dislocation nature (25).
Note that a Lagrangian density L() of translation gauge potentials  cannot be
built in the standard Yang-Mills form since the Lie algebra of the ane group does not
admit an invariant nondegenerate bilinear form. To construct L(), one can utilize the
torsion
F  = D  D 
of the linear connection with respect to the soldering form . The general form of a
Lagrangian density L() is given by the expression
L() = 21 [a1F  F + a2F F  + a3F F 
p
+a4 F  F     +    ] g:
This Lagrangian density diers from the familiar gravitational Lagrangian densities.
In particular, it contains the mass-like term originated from the Lagrangian density (23)
for displacement elds u under the gauge condition u = 0. Solutions of the corresponding
14
eld equations show that elds  make contribution to the standard gravitational eects.
In particular, they lead to the "Yukawa type" modication of Newton's gravitational
potential.

References
1. J. Cari~nena, M. Crampin and L. Ibort, Di. Geom. Appl., 1, 345 (1991).
2. G. Giachetta and L. Mangiarotti, Int. J. Theor. Phys., 29, 789 (1990).
3. M. Gotay, in Mechanics, Analysis and Geometry: 200 Years after Lagrange, ed.
M.Francaviglia (Elseiver Science Publishers B.V., 1991) p. 203.
4. C. Gunther, J. Di. Geom., 25, 23 (1987).
5. S. Hawking and G. Ellis, The Large Scale Structure of a Space-Time (Cambrifge
Univ. Press, Cambridge, 1973).
6. F. Hehl, J. McCrea, E. Mielke and Y. Ne'eman: Metric-ane gauge theory of
gravity, Physics Reports (1995) (to appear).
7. D. Ivanenko and G. Sardanashvily, Physics Reports, 94, 1 (1983).
8. A. Kadic and D. Edelen, A Gauge Theory of Dislocations and Disclinations
(Springer, New York, 1983).
9. I. Kolar, P.W. Michor, J. Slovak, Natural operations in dierential geometry,
(Springer-Verlag, Berlin etc. 1993).
10. B. Kupershmidt, Geometry of Jet Bundles and the Strucuture of Lagrangian and
Hamiltonian Formalisms, Lect. Notes in Math., 775, 162 (1980).
11. L. Mangiarotti and M. Modugno, in Geometry and Physics, ed. M.Modugno
(Pitagora Editrice, Bologna, 1982) p. 135.
12. E. Mielke: Geometrodynamics of Gauge Fields - On the Geometry of Yang-Mills
and Gravitational Gauge Theories (Akademic-Verlag, Berlin, 1987).
13. J. Ne'eman and Dj. Sijacki, Annals of Phsics, 120, 292 (1979).
14. R. Percacci, Geometry of Nonlinear Field Theories (World Scientic, Singapore,
1986).
15. R. Percacci, Nuclear Physics, B353, 271 (1991).

15
16. G.Sardanashvily, Acta Physica Polonica, B21, 583 (1990).
17. G. Sardanashvily, J. Math. Phys., 33, 1546 (1992).
18. G. Sardanashvily and O. Zakharov, Gauge Gravitation Theory (World Scientic,
Singapore, 1992).
19. G. Sardanashvily and O. Zakharov, Di. Geom. Appl., 3, 245 (1993).
20. G.Sardanashvily, Gauge Theory in Jet Manifolds (Hadronic Press, Palm Harbor,
1993)
21. G. Sardanashvily, Generalized Hamiltonian Formalism for Field Theory. Constraint
Systems (World Scientic, Singapore, 1994).
22. G. Sardanashvily, Multimomentum Hamiltonian Formalism in Field Theory, E-
print: hp-th/9403172, 9405040.
23. G. Sardanashvily, Gravitation Singularities of the Caustic Type, E-print: gr-
qc/9404024.
24. G. Sardanashvily, Gravity as a Higgs Field, E-print: gr-qc/9405013, 9407032.
25. G.Sardanashvily, Hamiltonian eld systems on composite manifolds, E-print: hep-
th/9409159.
26. R. Sulanke and P. Wintgen, Ddieretialgeometrie und Faserbundel ( Veb Deutsher
Verlag der Wissenschaften, Berlin, 1972).
27. D. Saunders, The Geometry of Jet Bundles, (Cambridge Univ. Press, Cambridge,
1989).

16