a

r
X
i
v
:
h
e
p
-
t
h
/
0
3
1
1
0
6
0
v
1


7

N
o
v

2
0
0
3
General Principles of Brane Kinematics and
Dynamics
Matej Pavsic
Jozef Stefan Institute, Jamova 39, 1000 Ljubljana, Slovenia
e-mail: [email protected]
February 1, 2008
Abstract
We consider branes as points in an innite dimensional brane
space Mwith a prescribed metric. Branes move along the geodesics of
M. For a particular choice of metric the equations of motion are equiv-
alent to the well known equations of the Dirac-Nambu-Goto branes
(including strings). Such theory describes free fall in M-space. In
the next step the metric of M-space is given the dynamical role and
a corresponding kinetic term is added to the action. So we obtain a
background independent brane theory: a space in which branes live is
M-space and it is not given in advance, but comes out as a solution to
the equations of motion. The embedding space (target space) is not
separately postulated. It is identied with the brane conguration.
1 Introduction
Theories of strings and higher dimensional extended objects, branes, are
very promising in explaining the origin and interrelationship of the funda-
mental interactions, including gravity. But there is a cloud. It is not clear
what is a geometric principle behind string and brane theories and how to
formulate them in a background independent way. An example of a back-
ground independent theory is general relativity where there is no preexisting
space in which the theory is formulated. The dynamics of the 4-dimensional
space (spacetime) itself results as a solution to the equations of motion.
The situation is sketched in Fig.1. A point particle traces a world line in
spacetime whose dynamics is governed by the Einstein-Hilbert action. A
closed string traces a world tube, but so far its has not been clear what is
the appropriate space and action for a background independent formulation
of string theory.
Here I will report about a formulation of string and brane theory (see
also ref. [1]) which is based on the innite dimensional brane space M.
The points of this space are branes and their coordinates are the brane
(embedding) functions. In M-space we can dene the distance, metric,
1
I[g

] =
_
d
4
x

g R
?
. . . . .
. . ... . . . . . . . . . .. . .. . . . .. . . . . . .. . . .. . .. ....
.......
........ ...... ..... . ...
..... .
Figure 1: To point particle there corresponds the Einstein-Hilbert action in space-
time. What is a corresponding space and action for a closed string?
connection, covariant derivative, curvature, etc. We show that the brane
dynamics can be derived from the principle of minimal length in M-space;
a brane follows a geodetic path in M. The situation is analogous to the free
fall of an ordinary point particle as described by general relativity. Instead
of keeping the metric xed, we then add to the action a kinetic term for
the metric of M-space and so we obtain a background independent brane
theory in which there is no preexisting space.
2 Brane space M (brane kinematics)
We will rst treat the brane kinematics, and only later we will introduce a
brane dynamics. We assume that the basic kinematically possible objects are
n-dimensional, arbitrarily deformable branes V
n
living in an N-dimensional
embedding (target) space V
N
. Tangential deformations are also allowed.
This is illustrated in Fig. 2. Imagine a rubber sheet on which we paint a
grid of lines. Then we deform the sheet in such a way that mathematically
the surface remains the same, nevertheless the deformed object is physically
dierent from the original object.
We represent V
n
by functions X

(
a
) , = 0, 1, ..., N1, where
a
, a =
0, 1, 2, ..., n1 are parameters on V
n
. According the assumed interpretation,
dierent functions X

(
a
) can represent physically dierent branes. That
is, if we perform an active dieomorphism
a

a
= f
a
(), then the new
functions X

(f
a
()) = X

() represent a physically dierent brane V

n
. For
a more complete and detailed discussion see ref. [1].
The set of all possible V
n
forms the brane space M. A brane V
n
can be
considered as a point in M parametrized by coordinates X

(
a
) X
()
which bear a discrete index and n continuous indices
a
. That is, () as
superscript or subscript denotes a single index which consists of the discrete
part and the continuous part ().
In analogy with the nite-dimensional case we can introduce the distance
2
Figure 2: Examples of tangentially deformed membranes. Mathematically the
surface on the left and is the same as the surface on the right. Physically the two
surfaces are dierent.
d in the innite-dimensional space M:
d
2
=
_
d d

(, ) dX

() dX

() =
()()
dX
()
dX
()
, (1)
where

(, )
()()
is the metric in M. Let us consider a particular
choice of metric

()()
=
_
|f| g

( ), (2)
where f det f
ab
is the determinant of the induced metric f
ab

a
X

b
X

on the sheet V
n
, whilst g

is the metric tensor of the em-

bedding space V
N
, and an arbitrary function of
a
or, in particular, a
constant. Then the line element (1) becomes
d
2
=
_
d
_
|f| g

dX

()dX

(). (3)
The invariant volume (measure) in M is
_
||DX = (Det

(, ))
1/2

,
dX

(). (4)
Here Det denotes a continuum determinant taken over , as well as over
, . In the case of the diagonal metric (2) we have
_
||DX =

,
_
_
|f| |g|
_
1/2
dX

() (5)
Tensor calculus in M-space is analogous to that in a nite dimensional
space. The dierential of coordinates dX

() dX
()
is a vector in M.
The coordinates X
()
can be transformed into new coordinates X

()
which
are functionals of X
()
:
X

()
= F
()
[X]. (6)
3
If functions X

() represent a brane V
n
, then functions X

() obtained
from X

() by a functional transformation represent the same (kinemati-

cally possible) brane.
Under a general coordinate transformation (6) a generic vector A
()

() transforms as
1
A
()
=
X

()
X
()
A
()

_
d
X

()
X

()
A

() , (7)
where /X

() denotes the functional derivative.

Similar transformations hold for a covariant vector A
()
, a tensor
B
()()
, etc.. Indices are lowered and raised, respectively, by
()()
and

()()
, the latter being the inverse metric tensor satisfying

()()

()()
=
()
()
. (8)
As can be done in a nite-dimensional space, we can also dene the
covariant derivative in M. When acting on a scalar A[X()] the covariant
derivative coincides with the ordinary functional derivative:
A
;()
=
A
X

()
A
,()
. (9)
But in general a geometric object in M is a tensor of arbitrary rank,
A

1
(
1
)
2
(
2
)...

1
(
1
)
2
(
2
)...
, which is a functional of X

(), and its covariant

derivative contains the anity
()
()()
composed of the metric
()(

)
[3].
For instance, when acting on a vector the covariant derivative gives
A
()
;()
= A
()
,()
+
()
()()
A
()
(10)
In a similar way we can write the covariant derivative acting on a tensor of
arbitrary rank.
In analogy to the notation as employed in the nite dimensional tensor
calculus we can use the following variants of notation for the ordinary and
covariant derivative:

()


X
()

()
for functional derivative
D
DX

()

D
DX
()
D
()
for covariant derivative in M (11)
Such shorthand notations for functional derivative is very eective.
1
A similar formalism, but for a specic type of the functional transformations, namely
the reparametrizations which functionally depend on string coordinates, was developed
by Bardakci [2]
4
3 Brane dynamics: brane theory as free fall in M-
space
So far we have considered kinematically possible branes as the points in the
brane space M. Instead of one brane we can consider a one parameter family
of branes X

(,
a
) X
()
(), i.e., a curve (or trajectory) in M. Every
trajectory is kinematically possible in principle. A particular dynamical
theory then selects which amongst those kinematically possible branes and
trajectories are also dynamically possible. We will assume that dynamically
possible trajectories are geodesics in M described by the minimal length
action [1]:
I[X
()
] =
_
d

)(

)

X
(

)

X
(

)
_
1/2
. (12)
Let us introduce the shorthand notation

(

)(

)

X
(

)

X
(

)
(13)
and vary the action (12) with respect to X
()
(). If the expression for the
metric
(

)(

)
does not contain the velocity

X

we obtain
1

1/2
d
d
_

X
()

1/2
_
+
()
(

)(

X
(

)

X
(

= 0 (14)
which is a straightforward generalization of the usual geodesic equation from
a nite-dimensional space to an innite-dimensional M-space.
Let us now consider a particular choice of the M-space metric:

)(

)
=
_
|f(

)|
_

X
2
(

)
(

(15)
where

X
2
g

is the square of velocity

X

. Therefore, the met-

ric (15) depends on velocity. If we insert it into the action (12), then after
performing the functional derivatives and the integrations over and
a
(im-
plied in the repeated indexes (

), (

)) we obtain the following equations

of motion:
d
d
_
1

1/2
_
|f|
_

X
2

_
+
1

1/2

a
_
_
|f|
_

X
2

a
X

_
= 0 (16)
If we take into account the relations
d
_
|f|
d
=

_
|f|
f
ab

f
ab
=
_
|f| f
ab

a

X

b
X

=
_
|f|
a
X

a

X

(17)
and

_

X
2

_

X
2
= 1
d
d
_

X

_

X
2
_

X

= 0 (18)
5
it is not dicult to nd that
d
d
= 0 (19)
Therefore, instead of (16) we can write
d
d
_
_
|f|
_

X
2

_
+
a
_
_
|f|
_

X
2

a
X

_
= 0. (20)
This are precisely the equation of motion for the Dirac-Nambu-Goto brane,
written in a particular gauge.
The action (12) is by denition invariant under reparametrizations of
a
.
In general, it is not invariant under reparametrization of the parameter .
If the expression for the metric
(

)(

)
does not contain the velocity

X

,
then the action (12) is invariant under reparametrizations of . This is no
longer true if
(

)(

)
contains

X

. Then the action (12) is not invariant

under reparametrizations of .
In particular, if metric is given by eq. (15), then the action becomes
explicitly
I[X

()] =
_
d
_
d
_
|f|
_

X
2
_
1/2
(21)
and the equations of motion (16), as we have seen, automatically contain
the relation
d
d
_

X
()

X
()
_

d
d
_
d
_
|f|
_

X
2
= 0. (22)
The latter relation is nothing but a gauge xing relation, where by gauge
we mean here a choice of parameter . The action (12), which in the case of
the metric (15) is not reparametrization invariant, contains the gauge xing
term.
In general the exponent in the Lagrangian is not necessarily
1
2
, but can
be arbitrary:
I[X
()
] =
_
d
_

)(

)

X
(

)

X
(

)
_
a
. (23)
For the metric (15) we have explicitly
I[X

()] =
_
d
_
d
_
|f|
_

X
2
_
a
(24)
The corresponding equations of motion are
d
d
_
a
a1

_
|f|
_

X
2

_
+ a
a1

a
_

_
|f|
_

X
2

a
X

_
= 0. (25)
We distinguish two cases:
6
(i) a = 1. Then the action is not invariant under reparametrizations of
. The equations of motion (25) for a = 1 imply the gauge xing relation
d/d = 0, that is, the relation (22).
(ii) a = 1. Then the action (24) is invariant under reparametrizations of
. The equations of motion for a = 1 contain no gauge xing term. In both
cases, (i) and (ii), we obtain the same equations of motion (20).
Let us focus our attention to the action with a = 1:
I[X
()
] =
_
d
_

)(

)

X
(

)

X
(

)
_
=
_
d d
_
|f|
_

X
2
(26)
It is invariant under the transformations

() (27)

a
=
a
(
a
) (28)
in which and
a
do not mix.
Invariance of the action (26) under reparametrizations (27) of the evolu-
tion parameter implies the existence of a constraint among the canonical
momenta p
()
and coordinates X
()
. Momenta are given by
p
()
=
L

X
()
= 2
()(

)

X
(

)
+

)(

X
()

X
(

)

X
(

)
=

_
|f|
_

X
2

. (29)
By distinguishing covariant and contravariant components one nds
p
()
=

X
()
=
()(

)

X
(

)
, p
()
=

X
()
. (30)
We dene p
()
p

() p

,

X
()

()

X

. Here p

and

X

have the meaning of the usual nite dimensional vectors whose components
are lowered and raised by the nite-dimensional metric tensor g

and its
inverse g

: p

= g

,

X

= g

.
The Hamiltonian belonging to the action (26) is
H = p
()

X
()
L =
_
d
_

X
2

_
|f|
(p

2
|f|) = p
()
p
()
K = 0 (31)
where K = K[X
()
] =
_
d
_
|f|
_

X
2
= L. It is identically zero. The

X
2
entering the integral for H is arbitrary due to arbitrary reparametriza-
tions of (which change

X
2
). Consequently, H vanishes when the following
expression under the integral vanishes:
p

2
|f| = 0 (32)
Expression (32) is the usual constraint for the Dirac-Nambu-Goto brane
(p-brane). It is satised at every
a
.
7
In ref. [1] it is shown that the constraint is conserved in and that as a
consequence we have
p

a
X

= 0. (33)
The latter equation is yet another set of constraints
2
which are satised at
any point
a
of the brane world manifold V
n+1
.
Both kinds of constraints are thus automatically implied by the action
(26) in which the choice (15) of M-space metric tensor has been taken.
Introducing a more compact notation
A
= (,
a
) and X
()
()
X

(
A
) X
()
we can write
I[X
()
] =
()(

)

X
()

X
(

)
=
_
d
n+1

_
|f|
_

X
2
(34)
where

)(

)
=
_
|f(

)|
_

X
2
(

)
(

)(

(35)
Variation of the action (34) with respect to X
()
gives
d

X
()
d
+
()
(

)(

X
(

)

X
(

)
= 0 (36)
which is the geodesic equation in the space M
V
n+1
of brane world manifolds
V
n+1
described by X
()
. For simplicity we will omit the subscript and call
the latter space M-space as well.
Once we have the constraints we can write the rst order or phace space
action
I[X

, p

, ,
a
] =
_
d d
_
p

2
_
|f|
(p

2
|f|)
a
p

a
X

_
,
(37)
where and
a
are Lagrange multipliers. It is classically equivalent to the
minimal surface action for the (n + 1)-dimensional world manifold V
n+1
I[X

] =
_
d
n+1
(det
A
X

B
X

)
1/2
. (38)
This is the conventional DiracNambuGoto action, invariant under
reparametrizations of
A
.
4 Dynamical metric eld in M-space
Let us now ascribe the dynamical role to the M-space metric. From M-
space perspective we have motion of a point particle in the presence of a
metric eld
()(

)
which is itself dynamical.
2
Something similar happens in canonical gravity. Moncrief and Teitelboim [4] have
shown that if one imposes the Hamiltonian constraint on the Hamilton functional then
the momentum constraints are automatically satised.
8
As a model let us consider the action
I[] =
_
DX
_
||
_

()(

)

X
()

X
(

)
+

16
R
_
. (39)
where is the determinant of the metric
()(

)
and a constant. Here R
is the Ricci scalar in M-space, dened according to R =
()(

)
R
()(

)
,
where R
()(

)
is the Ricci tensor in M-space [1].
Variation of the action (39) with respect to X
()
and
()(

)
leads
to (see ref.[1]) the geodesic equation (36) and to the Einstein equations in
M-space

X
()

X
()
+

16
R
()(

)
= 0 (40)
In fact, after performing the variation we had a term with Rand a term with

X
()

X
()
in the Einstein equations. But, after performing the contraction
with the metric, we nd that the two terms cancel each other resulting in
the simplied equations (40) (see ref.[1]).
The metric
()(

)
is a functional of the variables X
()
and in
eqs. (36),(40) we have a system of functional dierential equations which
determine the set of possible solutions for X
()
and
()(

)
. Our brane
model (including strings) is background independent: there is no preexisting
space with a preexisting metric, neither curved nor at.
We can imagine a model universe consisting of a single brane. Although
we started from a brane embedded in a higher dimensional nite space, we
have subsequently arrived at the action(39) in which the dynamical variables
X
()
and
()(

)
are dened in M-space. In the latter model the concept
of an underlying nite dimensional space, into which the brane is embedded,
is in fact abolished. We keep on talking about branes for convenience
reasons, but actually there is no embedding space in this model. The metric

()(

)
[X] is dened only on the brane. There is no metric of a space
into which the brane is embedded. Actually, there is no embedding space.
All what exists is a brane conguration X
()
and the corresponding metric

()(

)
in M-space.
A system of branes (a brane conguration) Instead of a single brane
we can consider a system of branes described by coordinates X
(,k)
, where
k is a discrete index that labels the branes (Fig. 3). If we replace () with
(, k), or, alternatively, if we interpret () to include the index k, then the
previous action (39) and equations of motion (36),(40) are also valid for a
system of branes.
A brane conguration is all what exists in such a model. It is identied
with the embedding space
3
.
3
Other authors also considered a class of brane theories in which the embedding space
has no prior existence, but is instead coded completely in the degrees of freedom that reside
on the branes. They take for granted that, as the background is not assumed to exist, there
are no embedding coordinates (see e.g., [5]). This seems to be consistent with our usage of
9
(
A
, k)
Figure 3: The system of branes is represented as being embedded in a nite-
dimensional space V
N
. The concept of a continuous embedding space is only an
approximation which, when there are many branes, becomes good at large scales
(i.e., at the macroscopic level). The metric is dened only at the points (, k)
situated on the branes. At large scales (or equivalently, when the branes are small
and densely packed together) the set of all the points (, k) is a good approximation
to a continuous metric space V
N
.
From M-space to spacetime We now dene M-space as the space of all
possible brane congurations. Each brane conguration is considered as a
point in M-space described by coordinates X
(,k)
. The metric
(,k)(

,k

)
determines the distance between two points belonging to two dierent brane
congurations:
d
2
=
(,k)(

,k

)
dX
(,k)
dX
(

,k

)
(41)
where
dX
(,k)
= X
(,k)
X
(,k)
. (42)
Let us now introduce another quantity which connects two dierent
points, in the usual sense of the word, within the same brane conguration:

(, k) X
(

,k

)
X
(,k)
. (43)
and dene
s
2
=
(,k)(

,k

(, k)

, k

). (44)
In the above formula summation over the repeated indices and is as-
sumed, but no integration over ,

and no summation over k, k

.
Eq.(44) denotes the distance between the points within a given brane
conguration. This is the quadratic form in the skeleton space S. The
metric in the skeleton space S is the prototype of the metric in target
space V
N
(the embedding space). A brane conguration is a skeleton S of a
target space V
N
.
X
()
which, at the fundamental level, are not considered as the embedding coordinates,
but as the M-space coordinates. Points of M-space are described by coordinates X
()
,
and the distance between the points is determined by the metric
()(

)
, which is
dynamical.. In the limit of innitely many densely packed branes, the set of points (
A
, k)
is supposed to become a continuous, nite dimensional metric space VN.
10
5 Conclusion
We have taken the brane space M seriously as an arena for physics. The
arena itself is also a part of the dynamical system, it is not prescribed in
advance. The theory is thus background independent. It is based on a
geometric principle which has its roots in the brane space M. We can thus
complete the picture that occurred in the introduction:
I[g

] =
_
d
4
x

g R
I[
()(

)
] =
_
DX
_
|| R
. . . . .
. . ... . . . . . . . . . .. . .. . . . .. . . . . . .. . . .. . .. ....
.......
........ ...... ..... . ...
..... .
Figure 4: Brane theory is formulated in M-space. The action is given in terms
of the M-space curvature scalar R.
We have formulated a theory in which an embedding space per se does
not exist, but is intimately connected to the existence of branes (including
strings). Without branes there is no embedding space. There is no preexist-
ing space and metric: they appear dynamically as solutions to the equations
of motion. Therefore the model is background independent.
All this was just an introduction into a generalized theory of branes.
Much more can be found in a book [1] where the description with a metric
tensor has been surpassed. Very promising is the description in terms of
the Cliord algebra equivalent of the tetrad which simplies calculations
signicantly. The relevance of the concept of Cliord space for physics is
discussed in refs. [1], [6][10]).
There are possible connections to other topics. The system, or conden-
sate of branes (which, in particular, may be so dense that the corresponding
points form a continuum), represents a reference system or reference uid
with respect to which the points of the target space are dened. Such a
system was postulated by DeWitt [11], and recently reconsidered by Rovelli
[12] in relation to the famous Einsteins hole argument according to which
the points of spacetime cannot be identied. The brane model presented
here can also be related to the Mach principle according to which the mo-
tion of matter at a given location is determined by the contribution of all
the matter in the universe and this provides an explanation for inertia (and
inertial mass). Such a situation is implemented in the model of a universe
consisting of a system of branes described by eqs. (36),(40): the motion of a
k-th brane, including its inertia (metric), is determined by the presence of
all the other branes.
11
Acknowledgement
This work has been supported by the Ministry of Education, Science and
Sport of Slovenia under the contract PO-0517.
References
[1] M. Pavsic, The Landscape of Theoretical Physics: A Global View. From
Point Particles to the Brane World and Beyond, in Search of a Unifying
Principle, Kluwer, Dordrecht, 2001.
[2] K. Bardakci, Nucl. Phys. B271 (1986) 561.
[3] M. Pavsic, Found. Phys. 26 (1996) 159; Nuov. Cim. A 110 (1997) 369.
[4] V. Moncrief and C. Teitelboim, Phys. Rev. D 6 (1972) 966.
[5] L. Smolin, Phys. Rev. D 62 (2000) 086001.
[6] C. Castro, J. Chaos Solitons Fractals 11 (2000) 1721, hep-th/9912113;
J. Chaos, Solitons and Fractals 12 (2001) 1585, physics/0011040.
[7] W.M. Pezzaglia Jr, A Cliord Algebra Multivector Reformulation of
Field Theory, Dissertation, University of California, Davis, 1983; Clas-
sication of Multivector Theories and Modication o f the Postulates of
Physics, gr-qc/9306006; Polydimensional Relativity, a Classical Gen-
eralization of the Automorphism Invariance Principle, gr-qc/9608052;
Physical Applications of a Generalized Cliord Calculus: Papapetrou
Equations and Metamorphic Curvature, gr-qc/9710027.
[8] M. Pavsic, Found. Phys. 31, 1185 (2001) 1185, hep-th/0011216; Found.
Phys. 33 (2003) 1277, gr-qc/0211085; Class. Quant. Grav. 20 (2003)
2697, gr-qc/0111092; C. Castro and M. Pavsic, Phys. Lett. B 539 (2002)
133, hep-th/0110079.
[9] D. Hestenes, Space-Time Algebra, Gordon and Breach, New York, 1966;
D. Hestenes Cliord Algebra to Geometric Calculus D. Reidel, Dor-
drecht, 1984.
[10] N. S. Mankoc Borstnik and H. B. Nielsen, J. Math. Phys. 43 (2002)
5782, hep-th/0111257; J. Math. Phys. 44 (2003) 4817, hep-th/0303224.
[11] B.S. DeWitt, The Quantization of Geometry, in: L. Witten, Gravita-
tion: An Introduction to Current Research, Wiley, New York, 1962.
[12] C. Rovelli, Classical and Quantum Gravity 8 (1991) 297; 8 (1991) 317.
12