r
X
i
v
:
0
7
1
0
.
3
0
7
9
v
3
[
m
a
t
h

p
h
]
1
5
N
o
v
2
0
0
8
Fedosov Quantization of LagrangeFinsler
and HamiltonCartan Spaces and Einstein
Gravity Lifts on (Co) Tangent Bundles
Mihai Anastasiei
anastas@uaic.ro
Sergiu.Vacaru@gmail.com ; http://www.scribd.com/people/view/1455460sergiu
1
MSC: 83C99, 53D55, 53B40, 53B35
PACS: 04.20.q, 02.40.k, 02.90.+g, 02.40.Yy
1 Introduction
To construct a quantum theory for a given classical model is usually under
stood that it is necessary to elaborate a quantization procedure adapted to
certain fundamental eld equations and corresponding Lagrangians/ Hamil
tonians and theirs symmetries, constraints and locality. In various attempts
to develop quantum versions of gravity and nonlinear physical theories, one
provided dierent computation schemes when it is supposed that all con
straints can be solved, gauge symmetries can be represented by shift sym
metries and there are certain canonical forms for the Poisson brackets. Nev
ertheless, such quantization usually destroy global symmetries, locality and
may result into violation of local Lorentz symmetry.
There were proposed dierent sophisticate constructions with formal and
partial solutions for quantum gravity and eld interactions theories. We
cite here the BRST quantization methods for nonAbelian and open gauge
algebras [1, 2, 3, 4], deformation quantization [5, 6, 7, 8], quantization of
general Lagrange structures and, in general, BRST quantization without
Lagrangians and Hamiltonians [9, 10], Wgeometry and Moyal deformations
of gravity via strings and branes [11, 12, 13] and quantum loops and spin
networks [14, 15, 16].
In the abovementioned approaches, it is necessary to quantize curved
phase spaces in a manner that is explicitly covariant on phasespace coordi
nates. Indeed, for instance, the key ingredients of Fedosov and BRST meth
ods, see [9, 10, 17, 18] and references therein, is the embedding of the system
into the cotangent bundle over its phase space. There were also elaborated
such models following standard geometric constructions over Riemannian
manifolds and formal schemes with symplectic groupoids and contravariant
connections [17, 19].
A rigorous geometric approach to deformation quantization of gravity,
gauge theories and geometric mechanics models with constraints and related
generalized LagrangeFinsler theories, see [20, 21, 22, 23, 24], shows that the
quantization schemes have to be developed for nonholonomic manifolds
1
and tangent and cotangent bundles endowed with nonlinear connection (N
connection) structure. The natural step in this direction is to apply the
1
i.e. manifolds endowed with nonholonomic (equivalent, anholonomic, or non
integrable distributions), see details and references in [25, 26, 27, 28, 29, 30]
2
methods of the geometry of Hamilton and Cartan spaces and generalizations
[31, 32] (such spaces are respectively dual to the Lagrange and Finsler spaces
and generalizations [33, 34]).
The aim of this work is to show how Karabegovs approach to Fedosov
deformation quantization [19, 35, 36] can be naturally extended for almost
K ahler manifolds endowed with canonical geometric structures generated
by semiRiemannian and/or Einstein metrics and LagrangeFinsler and
HamiltonCartan fundamental generating functions.
This paper is motivated by the following results: In Refs. [30, 21, 22,
37, 29], we concluded that classical and quantum gravity models on (co)
tangent bundles positively result in generalized Finsler like theories with vi
olation of local Lorentz symmetry. The conclusion was supported also by a
series of works on denition of spinors and eld interactions on (in general,
higher order) locally anisotropic spacetimes [38, 39], on low energy limits of
(super) string theory [40, 41, 42] and possible Finsler like phenomenological
implications and symmetry restriction of quantum gravity [43, 44, 45]. Here,
we emphasize that the nonholonomic quantum deformation formalism can
be redened for nonholonomic (pseudo) Riemannian, or RiemannCartan,
manifolds with bred structure. Such nonholonomic spaces, under well de
ned conditions and for corresponding LagrangeFinsler variables, possess
local Lorentz invariance for classical theories and seem to preserve it for
nonholonomic quantum deformations, see results from [23, 24].
The work is organized as follows: In Section 2 we recall the basis of
the LagrangeFinsler and HamiltonCartan geometry, nonholonomic lifts of
Einstein metrics on cotangent bundles and almost K ahler models of such
spaces. We also collect there some geometric constructions that we need
for further considerations. Section 3 is devoted to the formalism of canon
ical nonlinear connections and distinguished connections induced by La
grange and Hamilton fundamental functions. We introduce the concept of
HamiltonFedosov spaces and dene the corresponding almost symplectic
structures. In Section 4 we consider a generalization of the concept of con
nection to that of connectionpair resulting in FedosovHamilton operators
pairs which is necessary for denition of deformation quantization models
being invariant under symplectic morphisms and Legendre transform. We
provide Fedosovs theorems for connectionpairs which allows us to develop
an approach to geometric quantization of Hamilton and Einstein (general
ized on cotangent bundles) spaces in Section 5. We speculate on possible
quantum generalizations of gravitational eld equations on phase spaces and
their deformation quantization. Finally, in Section 6 we conclude the results.
3
2 LagrangeFinsler and HamiltonCartan Geom
etry and Einstein Spaces
In this section, we outline some results from the geometry of Lagrange
Finsler [33, 34] and HamiltonCartan [31, 32] spaces.
2.1 Canonical geometric objects on Lagrange and Hamilton
spaces
Let us consider a real suciently smooth manifold M, dimM = n 2. We
label the local coordinates x = {x
i
}, with base indices i, j, ... = 1, 2, ...n,
and write by TM and T
: T
= (x
i
, y
a
)}, for ber indices a, b, ... = n + 1, ..., n + n on
any local cart U TM, and
u = (x, p) = {
= (x
i
, p
a
)} on any local cart
U T
M, when p
a
are dual to y
a
. In order to apply the Einstein summation
rule for contracting base and ber indices, we shall use identications of type
y
i
y
n+i
, i.e. we suppose that indices like a, b, ... can split into respective
n +i, n +j.
Denition 2.1 A generalized Hamilton space is dened by a pair GH
n
=
(M, g
ij
(x, p)), where g
ij
(x, p) is a contravariant symmetric tensor eld, non
degenerate and of constant signature on
T
M = T
M.
We note that a contravariant tensor of type
g
ij
(x, p) = e
i
i
(x, p)e
j
j
(x, p)g
i
(x) (1)
includes, for some vielbein elds e
i
i
j
(x) on M. In this paper, we consider that any
classical solution of the Einstein equations denes a spacetime manifold
M and a corresponding deformation quantization procedure on T
M de
nes the values e
i
i
(x) and g
ij
(x, p)
and discuss if there are any important particular properties for Einstein
manifolds (spaces) when g
i
j
(x) is subjected to the condition to solve on M
the gravitational eld equations with nonzero cosmological constant ,
R
i
j
(x) = g
i
j
(x), (2)
4
where
R
i
j
(x) is the Ricci tensor for the LeviCivita connection =
{
j
.
For simplicity, we shall work with a more particular class of spaces when
g
ij
(x, p) is dened by a Hamilton function H(x, p) :
Denition 2.2 A Hamilton space H
n
= (M, H(x, p)) is dened by a func
tion T
: T
g
ab
(x, p) =
2
H
p
a
p
b
(3)
is nondegenerate and of constant signature on
T
M.
Let L(x, y) be a regular dierentiable Lagrangian on U TM, with
nondegenerate Hessian (equivalently, fundamental tensor eld)
g
ab
(x, y) =
2
L
y
a
y
b
. (4)
Denition 2.3 A Lagrange space L
n
= (M, L(x, y)) is dened by a function
TM (x, y) L(x, p) R, i.e. a fundamental Lagrange function, which is
dierentiable on
TM and continuous on the null section of : TM M
and such that the (Hessian) tensor eld g
ab
(x, y) (4) is nondegenerate and
of constant signature on
TM.
We can dene the Legendre transform L H,
H(x, p) = p
a
y
a
L(x, y), (5)
where y = {y
a
} are solutions of the equations p
a
= L(x, y)/y
a
, and
(inversely) the Legendre transform H L,
L(x, y) = p
a
y
a
H(x, p), (6)
where p = {p
a
} is the solution of the equations y
a
= H(x, p)/p
a
.
2
2
In some monographs (for instance, see [34, 31]), it is considered the factor 1/2 in the
right sides of (3) and (4). We emphasize that in this paper (for simplicity) the Hamilton
and Lagrange functions will be supposed to be regular and related mutually by Legendre
transforms.
5
Following terminology from [31], we say that N
a
i
(14 ) and
N
ij
(15)
are Ldual if L and H are related by Legendre transform (5), or (6). In the
following constructions, we shall consider that to Legendre transform there
are associated the dieomorphisms
: TM U
U T
M, (x
i
, y
a
)
_
x
i
, p
a
=
L(x, y)
y
a
_
and
: T
M
U U TM, (x
i
, p
a
)
_
x
i
, y
a
=
H(x, p)
p
a
_
,
allowing to dene respectively pullback and pushforward of geometric ob
jects (functions, vectors, dierential forms, connections, tensors...) from
U to U and from U to
U, i.e. we dene Ldual geometric objects. For
instance, for a dierentiable function
1
f on U, we dene a dierentiable
function
1
f
1
f =
1
f
1
on
U and (inversely) for a dierentiable
function
2
f on
U, we have a dierentiable
_
2
f
_
0
2
f =
2
f
1
on U. Note that H = L
1
and L = H
1
. Similarly, for any vector
eld X on U, we get a vector eld
X T X
1
= T
1
X
on
U and (inversely) for a vector eld
X on
U, we get a vector eld
X T X
1
= T
1
X on U, where, for example, T is the
tangent map to . Dualizing the vector constructions, we obtain that for
any 1form on U, there is 1form
(T)
1
= (T
1
)
on
U and (inversely) for any form
on
U, we can consider
(T)
1
= (T
1
)
on U, where, for example, (T)
M,
respectively.
Denition 2.4 Any Whitney sums
TTM = hTM vTM (7)
and
TT
M = hT
M vT
M (8)
dene respectively nonlinear connection (Nconnection) structures paramet
rized by the local vector elds
e
i
=
x
i
N
a
i
(x, y)
y
a
on TM
6
and
e
i
=
x
i
+
N
ia
(x, p)
p
a
on T
M.
One says that a Nconnection denes on TM, or T
M, a conventional
horizontal (h) and vertical (v) splitting (decomposition).
Let consider a regular curve c() with real parameter , when c :
[0, 1] x
i
() U. It can be lifted to
1
(U)
TM as c() : [0, 1]
_
x
i
(), y
i
() =
dx
i
d
_
since the vector eld
dx
i
d
does not vanish on
TM. Fol
lowing techniques from variational calculus, one proves:
Theorem 2.1 The EulerLagrange equations,
d
d
L
y
i
L
x
i
= 0, (9)
are equivalent to the HamiltonJacobi equations,
dx
i
d
=
H
p
i
and
dp
i
d
=
H
x
i
, (10)
and to the nonlinear geodesic (semispray) equations
d
2
x
i
d
2
+ 2G
i
(x, y) = 0, (11)
where
G
i
=
1
2
g
ij
_
2
L
y
j
x
k
y
k
L
x
j
_
,
for g
ij
being the inverse to g
ij
(4).
Let us consider on T
M:
X
H
=
H
p
i
x
i
H
x
i
p
i
by the equation
i
X
H
= dH,
where i
X
H
denotes the interior product by X
H
. The same holds for any
function on T
M.
By Theorem 2.1 one has:
7
Corollary 2.1 The HamiltonJacobi equations (10) are equivalent to
dx
i
d
= {H, x
i
} and
dp
a
d
= {H, p
a
},
where the Poisson structure is dened by brackets
{
1
f,
2
f} = (X 1
f
, X 2
f
) (13)
for any functions
1
f(x, p) and
2
f(x, p) on T
M.
Proof. It can be obtained by a standard calculus in geometric mechan
ics.
The following theorem holds:
Theorem 2.2 The are canonical Nconnections dened respectively by reg
ular Lagrange L(x, y) and/ or Hamilton H(x, p) fundamental functions:
N
a
i
G
a
y
i
(14)
and
N
ij
1
2
_
{
g
ij
, H}
2
H
p
k
x
(i
g
j)k
_
, (15)
where,
g
ij
is the inverse to
g
ij
(3) and, for instance, a
(ij)
= a
ij
+a
ji
denotes
symmetrization of indices.
Proof. We can verify respectively that on any open sets U TM and
U
= (e
i
=
x
i
N
a
i
y
a
, e
b
=
y
b
), on TM, (16)
= (
e
i
=
x
i
+
N
ia
p
a
,
e
b
=
p
b
), on T
M, (17)
and their dual (coframe) structures
e
= (e
i
= dx
i
, e
b
= dy
b
+N
b
i
dx
i
), on (TM)
, (18)
= (
e
i
= dx
i
,
e
p
= dp
b
N
ib
dx
i
), on (T
M)
, (19)
when e
and
e
M)
in Nadapted form [equivalently, in distinguished form, in brief, dform]
if they are given by coecients dened with respect to frames e
(16) and
coframes e
(17)
and coframes
e
g =
g
=
g
ij
(x, p)e
i
e
j
+
g
ab
(x, p)
e
a
e
b
, (20)
on T
M, where
g
ij
is inverse to
g
ab
, and
g = g
= g
ij
(x, y)e
i
e
j
+g
ab
(x, y)e
a
e
b
,
on TM, where g
ij
is stated by g
ab
following g
ij
= g
n+i n+j
.
The following proposition holds:
Proposition 2.2 The canonical Nconnections N (14) and
N (15) dene
respectively the canonical almost complex structures J, on TM, and
J, on
T
M.
9
Proof. On TM one introduces the linear operator J acting on e
=
(e
i
, e
b
) (16) as follows:
J(e
i
) = e
n+i
and J(e
n+i
) = e
i
.
It is clear that J denes globally an almost complex structure (J J = I
for I being the unity matrix) on TM completely determined for Lagrange
spaces by a L(x, y). Now we provide the proof for T
M. Let us introduce a
linear operator
J acting on
e
= (
e
i
,
e
b
) (17) following formulas
J(
e
i
) = g
ia
e
n+i
and
J(
e
n+i
) =
e
i
.
Then
J denes globally an almost complex structure (
J
J = I
for I being the unity matrix) on T
J
(X, Y) = [X, Y] + [
JX,
JY]
J[
JX, Y]
J[X,
JY],
J
(X, Y) = [X, Y] + [JX, JY] J[JX, Y] J[X, JY], (21)
for any dvectors X and Y.
Hereafter, for simplicity and if one shall not result in ambiguities, we
shall present the Nadapted component formulas for geometric objects on
T
ija
:
ija
=
N
ia
x
j
N
ja
x
i
+
N
ib
N
ja
p
b
N
jb
N
ia
p
b
. (22)
They dene the coecients of the Nconnection curvature. One gets a
complex structure i.e
J
= 0 under some quite complicated conditions
on g
ab
(x, p) and N
ia
which will be not written here.
It should be noted here that the Nadapted (co) bases (16)(19) are
nonholonomic with nontrivial anholonomy coecients. For instance,
[e
, e
] = e
= W
(23)
with (antisymmetric) anholonomy coecients W
b
ia
=
a
N
b
i
and W
a
ji
=
a
ij
,
with :
a
ij
=
N
a
i
x
j
N
a
j
x
i
+N
b
i
N
a
j
p
b
N
b
j
N
a
i
p
b
.
10
2.2 Almost Kahler LagrangeHamilton structures
We can adapt to Nconnections various geometric structures on TM and
T
M a nonde
generate Nadapted 2form
=
1
2
(u)
e
.
The following Proposition holds:
Proposition 2.3 For any
on T
N = {
N
ia
} satisfying the conditions:
= (h
X, v
Y) = 0 and
.
= h
+v
, (24)
for any
X = h
X + v
X,
Y = h
Y + v
Y and h
(
X,
Y)
(h
X,h
Y), v
(
X,
Y)
(v
X,v
Y).
Proof. For
X =
e
= (
e
i
,
e
a
) and
Y =
e
= (
e
l
,
e
b
), where
=
(
e
i
,
e
a
) =
(
x
i
,
p
a
)
N
ib
(
p
b
,
p
a
) = 0.
Such conditions uniquely dene
N
ib
because
is nondegenerate, that is
rank
(
p
b
,
pa
) = n. Setting locally
=
1
2
ij
(u)e
i
e
j
+
1
2
ab
(u)
e
a
e
b
, (25)
where the rst term is for h
and the second term is v
, we get the second
formula in (24). Finally, we note that in this proposition the constructed
Nconnection
N
ib
, in general, is not a canonical one (15).
In a similar form, as in Proposition 2.3, we can construct a unique N
connection N = { N
a
i
} for any almost symplectic structure on TM (from
formal point of view, we have to omit in formulas the symbols * and
and use variables y
a
instead of p
a
).
A Nconnection
N (8) denes a unique decomposition of a dvector
X =
X
h
+
X
v
on T
M, for
X
h
= h
X and
X
v
= v
X, where the
11
projectors h and v denes respectively the distributions
N and
V. They
have the properties
h +v = I, h
2
= h, v
2
= v, h v = v h = 0.
This allows us to introduce on T
P I 2v = 2h I
acting on
e
= (
e
i
,
e
b
) (17) following formulas
P(
e
i
) =
e
i
and
P(
e
b
) =
e
b
.
In a similar form, a Nconnection N(7) induces an almost product structure
P on TM. One uses also the almost tangent operators
J(e
i
) = e
n+i
and J (e
a
) = 0, or J =
y
i
dx
i
;
J(
e
i
) =
g
ib
e
b
and
J
_
e
b
_
= 0, or J =
g
ia
p
a
dx
i
.
The operators
P,
J and
J are respectively Ldual to P, J and J if and
only if
N and N are Ldual.
For the aboveintroduced almost complex and almost product operators,
it is straightforward to prove
Proposition 2.4 Let (N,
N) be a pair of Ldual Nconnections. Then,
we can construct canonical dtensor elds (dened respectively by L(x, y)
and H(x, p) related by Legendre transforms (5) and/or (6)) :
J =
a
i
e
a
e
i
+
i
a
e
i
e
a
,
J =
g
ia
e
a
e
i
+
g
ia
e
i
e
a
corresponding to the Ldual pair of almost complex structures (J,
J) ,
P = e
i
e
i
e
a
e
a
,
P =
e
i
e
i
e
a
e
a
corresponding to the Ldual pair of almost product structures (P,
P) , and
almost symplectic structures
= g
aj
(x, y)e
a
e
i
and
=
a
i
e
a
e
i
(26)
12
Let us consider an important example:
A Finsler manifold (space) F
n
= (M, F(x, y)) is a particular case of
Lagrange space, when the regular Lagrangian L = F
2
is dened by a fun
damental Finsler function F(x, y) satisfying the conditions: 1. the positive
function F is dierentiable function on
TM continuous on the null section of
projection : TM M, 2. F(x, y) = F(x, y), i.e. it is 1homogeneous
on the bres of TM, and 3. the Hessian (4) dened in this case by F
2
is
positively dened on
TM.
It is used also the notion of Cartan space C
n
= (M, C(x, p)) for H =
C
2
(x, p) as a particular (1homogeneous on ber coordinates) case of Hamil
ton space when C satises the same conditions as F but with respect to
coordinates p
a
(in brief, we can say that Cartan spaces are Finsler spaces on
T
M
(or tangent bundle TM) equipped with a Nconnection structure
N (or
N) is dened by a triple
H
2n
= (T
M,
,
J), where
(
X,
Y)
g (
JX, Y) (or H
2n
= (TM, , J), where (X, Y) g (JX, Y)). A space
H
2n
is almost K ahler, denoted
K
2n
if d
= 0.
The following theorem holds:
Theorem 2.3 The Lagrange and Hamilton spaces can be represented as al
most K ahler spaces on, respectively, on TM and T
M of canonical 1
forms, respectively, dened by a regular Lagrangian L and Hamiltonian H
related by a Legendre transform,
=
L
y
i
e
i
and
= p
i
dx
i
,
for which
= d and
=d
,
13
see (12). As a result, we get that d = 0 and d
= 0, which correspond
to the Denition 2.8.
In this paper, we shall work with almost K ahler models on (co) tan
gent bundles dened canonically by (pseudo) Riemannian metrics on base
manifolds, see (1), and/or (eective, or for regular mechanics) Lagrangians
(Hamiltonians). Finally, we emphasize that realistic classical and quantum
models are elaborated in explicit form for some classes of linear connections
dened to satisfy certain physical principles and constructed geometrically
to be adapted, or not, to a Nconnection structure. We shall perform such
classical and quantum constructions in the following sections.
3 Nonlinear Connections and Almost Symplectic
Geometry
In this section, we consider the almost symplectic geometry induced by reg
ular Hamiltonians and corresponding canonical Nconnections dened natu
rally, for gravitational and/or geometric mechanics models, on (co) tangent
bundles.
3.1 Canonical Nconnections and dconnections for Lagran
ge and Hamilton spaces
Let D be a linear connection on TM when for a Lduality between the
tangent and corresponding cotangent bundles there are dened pullback
and pushforward maps. We can dene a linear connection
D on T
M as
follows:
D
X
Y (D
X
Y )
,
for any vector elds
X and
Y on T
D on T
D
X
Y (
D
X
Y )
,
for any vector elds X and Y on TM.
Denition 3.1 A linear connection D or (
D) on TM (or T
M) is a
distinguished connection (dconnection) if it is compatible with the almost
product structure DP = 0 (or
D
P = 0).
14
For Ldual Lagrange and Hamilton spaces, one follows that DP = 0
induces
D
P = 0, and inversely. The coecients of dconnections can be
dened with respect to Nadapted frames,
D
e
and D
e
D
e
k
e
j
L
i
jk
e
i
,
D
e
k
e
b
L
b
a k
e
a
,
D
e
c
e
j
C
i c
j
e
i
,
D
e
c
e
b
C
bc
a
e
a
,
when
= {L
i
jk
,
L
a
bk
,
C
i
jc
, C
a
bc
} and
= {
L
i
jk
,
L
b
a k
,
C
i c
j
,
C
bc
a
}
dene corresponding h and vsplitting of covariant derivatives
D =(
h
D,
v
D) and
D =(
h
D,
v
D) ,
where
h
D = {L
i
jk
,
L
a
bk
},
v
D = {
C
i
jc
, C
a
bc
} and
h
D = {
L
i
jk
,
L
b
a k
},
v
D = {
C
i c
j
,
C
bc
a
}.
We shall work with a more special class of dconnections:
Denition 3.2 A linear connection
n
D (or
n
D) on TM (or T
M) is N
linear if it preserves under parallelism the Whitney sum N (7) (or
N (8))
i.e it is a dconnection and is compatible with the almost tangent structure
J (or
J) i.e.
n
DJ = 0 (or
n
D
J = 0).
This is a class of Nlinear connections characterized by the property:
Proposition 3.1 The Nadapted coecients of Nlinear connections
n
D {
n
= (L
i
jk
,
C
i
jc
)} and
n
D {
n
= (
L
i
jk
,
C
i c
j
)} of
respective Ldual Lagrange and Hamilton spaces are:
n
= {L
i
jk
,
L
n+i
n+jk
= L
i
jk
,
C
i
jc
, C
n+i
n+jc
=
C
i
jc
} (27)
and
n
= {
L
i
jk
,
L
n+i
n+j k
=
L
i
jk
,
C
i c
j
,
C
n+i c
n+j
=
C
i c
j
}. (28)
15
Proof. By a straightforward computation for coecients (27) and (28),
when, for instance,
n
D
= (
h
D
i
,
v
D
a
), for
h
D = {
L
i
jk
} and
v
D =
{
C
i c
j
}, we can verify that the conditions considered in Denition 3.2 are
satised.
The connection 1form of Nlinear connection
n
=
n
=
{
n
i
j
} is dened by
n
i
j
=
L
i
jk
e
k
+
C
i c
j
e
c
, (29)
where the vcomponents
n
a
b
are identied with the hcomponents
n
i
j
following formulas
n
n+i
n+j
=
n
a
b
. The following theorem holds:
Theorem 3.1 On a Hamilton space, the structure equations for
n
i
j
are
d
e
k
e
j
k
j
=
T
k
,
d
e
c
+
e
a
a
c
=
T
c
,
d
n
i
j
n
i
k
k
j
=
R
i
j
,
where the 2form of torsion T
= {T
k
, T
c
} is computed
T
k
=
1
2
T
k
ij
e
i
e
j
+
C
k c
j
e
j
e
c
,
T
a
=
1
2
ija
e
i
e
j
+
1
2
P
aic
e
i
e
c
+
1
2
S
bc
a
e
b
e
c
,
for the Nconnection curvature
ija
(22) and
T
k
ij
=
L
k
ij
L
k
ji
,
S
bc
a
=
C
b c
a
C
c b
a
,
P
aic
=
g
ae
(
L
e
ic
e
e
(
N
ic
)) , (30)
and the 2form of curvature
R
=
_
R
i
j
,
R
a
b
_
, with
R
i
j
=
R
n+i
n+j
,
is computed
R
i
j
=
1
2
R
i
jkm
e
k
e
m
+
P
i c
jk
e
k
e
c
+
1
2
S
i bc
j
e
b
e
c
,
where
R
i
jkm
=
e
m
(
L
i
jk
)
e
k
(
L
i
jm
) (31)
+
L
o
jk
L
i
om
L
o
jm
L
i
ok
+
C
i o
j
oka
,
P
i c
jk
=
e
c
(
L
i
jk
)
h
D
k
(
C
i c
j
) +
C
i o
j
P
c
ko
,
S
i bc
j
=
e
c
(
C
i b
j
)
e
b
(
C
i c
j
) +
C
k b
j
C
i c
k
C
k c
j
C
i b
k
,
for nonboldface
e
c
= /p
c
.
16
Proof. It is a straightforward dierential computation for 1form (29).
M completely de
ned, respectively, by the fundamental Lagrange (see [34] and, for applica
tions to geometric quantization, [21, 24]) and Hamilton functions.
Theorem 3.2 There exists a canonical Nlinear connection
D {
= (
L
i
jk
,
C
i c
j
)} on a Hamilton space H
n
= (M, H(x, p))
endowed with canonical Nconnection
N
ij
(15) satisfying the conditions:
1)
h
D
i
g
kj
= 0 and
v
D
a
g
kj
= 0, 2)
T
k
ij
= 0 and
S
bc
a
= 0 and 3)
L
i
jk
=
1
2
g
is
(
e
j
(
g
sk
) +
e
k
(
g
js
)
e
s
(
g
jk
)) ,
C
i c
j
=
1
2
g
js
e
c
(
g
si
) (32)
dened with respect to Nadapted bases
e
(19)
dened by
N
ij
(15). By a direct computation, we can verify that the
conditions of this Theorem are satised only for such coecients and their
coordinate/frame transform.
The coecients (32) are just the Christoel symbols on (co) tangent
space T
D is an almost symplectic d
connection satisfying the conditions
D
= 0 and
D
J = 0 (33)
17
and being completely dened by a Hamiltonian H(x, p) for
(
X, .)
g (
JX, .) .
Proof. A Hamilton space can be equivalently transformed into almost
K ahler space
K
2n
, see Denition 2.8. Considering the dtensor elds
associated to
and
J, see Proposition 2.4, and a covariant Nadapted
calculus dened by the canonical dconnection coecients (32) we can ver
ify that the almost symplectic structure compatibility conditions (33) are
satised.
In Ref. [20], it was introduced the concept of LagrangeFedosov manifold
as nonholonomic manifold with the Nconnection and almost symplectic
structure dened by a fundamental (in genera, eective) Lagrange function
L(x, y). On cotangent bundles, we can consider
Denition 3.3 A HamiltonFedosov space is a cotangent bundle endowed
with canonical Nconnection and the almost K ahler structure induced by a
fundamental Hamilton function H(x, p).
There are HamiltonFedosov spaces dened completely by a lift (1) of a
(pseudo) Riemannian metric on base M.
Theorem 3.3 Any Einstein manifold associated to a solution of (2), for a
lift (1) on T
can be
i
i
(
K
2n
)[[v]] of formal series in the variable v with coecients from
C
(
K
2n
) on a almost Poisson manifold (
K
2n
, {, }), see the almost sym
plectic form
and the Poisson brackets (13). Using the associative algebra
structure on C
(
K
2n
)[[v]] with a vlinear and vadically continuous star
product
1
f
2
f =
r=0
r
C(
1
f,
2
f) v
r
, (34)
19
where
r
C, r 0, are bilinear operators on C
(
K
2n
) with
0
C(
1
f,
2
f) =
1
f
2
f and
1
C(
1
f,
2
f)
1
C(
2
f,
1
f) = i{
1
f,
2
f}; for i being the complex
unity, we can construct a formal Wick product
1
a
2
a (z) exp
_
_
i
v
2
2
z
[1]
_
_
1
a(z)
2
a(z
[1]
) 
z=z
[1]
, (35)
for two elements a and b dened by series of the type
a(v, z) =
r0,{}0
a
r,{}
(u)z
{}
v
r
, (36)
where by {} we label a multiindex and
i
g
, where
is the inverse
to dtensor (20). This denes a formal Wick algebra
W
u
associated with
the tangent space T
u
K
2n
, for u
K
2n
, where the local coordinates on
K
2n
are parameterized in the form u = {u
K
2n
are labelled (u, z) = (u
, z
), where z
,
e
,
when deg
v
(v) = 1, deg
s
(z
) = 1, deg
a
(
e
,
e
D {
= (
L
i
jk
,
C
i c
j
)} with
coecients (32) can be extended to an operator
D(a )
_
e
(a) u
z
e
(a)
_
(
e
) +a d, (37)
on
W, where the Nadapted basis
z
e
is
e
redened in zvariables.
This canonical almost symplectic dconnection
D is a Nadapted deg
a
1
on
W,
are dened
(a) =
e
z
e
(a), and
1
(a) =
_
i
p+q
z
e
(a), if p +q > 0,
0, if p = q = 0,
(38)
20
where any a
W is homogeneous w.r.t. the grading deg
s
and deg
a
with deg
s
(a) = p and deg
a
(a) = q.
The doperators (38) dene the formula a = (
1
+
1
+
)(a), where a (a) is the projection on the (deg
s
, deg
a
)bihomogeneous
part of a of degree zero, deg
s
(a) = deg
a
(a) = 0;
is also a deg
a
graded
derivation of the dalgebra (
W , ) .
4.2 FedosovHamilton operatorpairs
Having dened doperators (38), we can perform a Fedosov type quanti
zation of Hamilton spaces in a Ldual form to Lagrange spaces [21, 24].
Nevertheless, such constructions would not reect completely the symplec
tic properties of Hamilton spaces and their quantum deformations. For a
cotangent bundle
: T
M M, a Nconnection
N (15) is a supple
mentary distribution hT
M = ker
,
where
M = ker(I +
P). If f Diff(T
on T
M ker(I
) is supple
mentary to vT
M ker(I +
) dene the
nonholonomic splitting (Whitney sum)
TT
M = hT
M wT
M. (39)
We can consider preferred
adapted frame and coframe structures
induced by the coecients of Nconnection:
Proposition 4.1 There are a canonical connectionpair
and associated
frames and coframes:
=
_
e
i
=
x
i
+
N
ia
p
a
,
w
e
b
=
p
b
g
bi
e
i
_
(40)
=
_
e
i
= dx
i
+
g
bi w
e
b
,
w
e
b
= dp
b
N
ib
dx
i
_
, (41)
where
g
ab
(3) and
N
ia
(15) are generated by a Hamilton fundamental func
tion H(x, p).
21
Proof. The proof is similar to that for Proposition 2.1 but adapted both
the the splitting (39), with respective h and vprojections, 2h
= I +
and
2w = I
, and to the splitting (8), with respective h and vprojections,
2h = I +
N and 2v = I
N.
We note that for Hamilton spaces a connectionpair structure
is
symmetric, i.e.
N
ia
=
N
ai
and
g
ab
=
g
ba
. For simplicity, in this work,
we shall restrict our considerations only to
symmetric congurations.
In order to perform geometric constructions adapted both to the N
connection and almost symplectic structure, it is necessary to work with
adapted bases (40) and (41) instead of, respectively, (17) and (19). For
instance, every vector eld X has two components
X = hX +vX = h
X +wX,
where wX = w(X). This denes the class of
tensor elds alternatively
to that of dtensor elds considered in the previous sections.
Denition 4.3 A linear connection
D on T
M is a
connection if
D
= 0 and
D w = 0.
We can characterize
D = (
h
D= {
L
i
jk
},
w
D= {
C
j k
i
}) by the
coecients computed with respect to
adapted frames (40) and (41):
D
e
k
e
j
=
L
i
jk
e
j
,
D
e
k
w
e
b
=
L
b
ck
w
e
c
,
D w
e
a
w
e
b
=
C
b a
c
w
e
c
,
D w
e
a
e
j
=
C
k a
j
e
k
,
for
L
n+i
n+j k
=
L
i
jk
and
C
n+k a
n+j
=
C
k a
j
.
Theorem 4.1 There is a canonical
connection
D = (
L
i
jk
,
C
k a
j
)
on T
D(
g
ab
) = 0 and
v
D(
g
ab
)=0,
T
k
ij
= 0 and
S
bc
a
= 0.
Proof. The torsion and curvature of
L
i
jk
=
1
2
g
km
(
e
i
g
mj
+
e
j
g
im
e
m
g
ij
)
1
2
g
km
(A
jim
+A
ijm
+A
mij
), (42)
C
j k
i
=
1
2
g
im
(
w
e
j
g
mk
+
w
e
k
g
im
+
w
e
m
g
jk
)
1
2
g
im
(B
jmk
+B
kmj
+B
mkj
),
where
A
ijm
=
ijm
and B
mkj
=
g
mi
(
g
ka w
e
j
N
ia
g
ja w
e
k
N
ia
),
the conditions of Theorem are satised.
The dieomorphism symmetry of Hamilton mechanical models and pos
sible lifts of Einstein spaces on cotangent bundles is an important character
istic of classical theories. In deformation quantization models, this property
can be preserved for regular Nconnection structures:
Denition 4.4 A dieomorphism f Diff(T
M) is called
Nregular
if the restriction of tangent map (f)
: hT
M TM to hT
M is a
dieomorphism.
For a connectionpair
, the concept of
Nregularity imposes the
equivalence of statements: 1) the pushforward map of
by f, dened as
= f
(
)f
1
, is a connectionpair and 2) f is
Nregular. This follows
from the mapping
: ker(I
(ker(I
)) =
ker(I
). The Nconnection
N associated to
is the pushforward
of
N by f. One says that
and
are frelated.
Following a calculus with local coordinate transforms, one proves:
Corollary 4.1 1) For
e
= (
e
i
,
w
e
b
) in (x, p) and
e
= (
e
i
,
w
e
b
) in
(x(x, p), p(x, p)) induced respectively by
and
, the following formulas
hold
f
(
e
i
) =
k
i
e
k
and f
(
w
e
b
) =
b
c
w
e
c
,
where
k
i
N
ka
=
e
i
(p
a
) and
b
c
=
w
e
b
(p
c
)
w
e
b
(x
k
)
N
kc
.
2) The pushforward of a Nconnection
N by a
Nregular dieomor
phism is a Nconnection if f is ber preserving, i.e. locally f(x, p) =
(x(x), p(x, p)).
23
Symplectic morphisms are dieomorphisms which transform a symplec
tic form
= dp
i
dx
i
into a symplectic form
= f
(
) = dp
i
dx
i
. By
coordinate transforms, one proves:
Theorem 4.2 For a f Diff(T
M) being
Nregular and satisfying the
condition f
(
)(
e
i
,
w
e
b
) =
b
i
, two from the next statements implies
the third: 1) f is a symplectic morphism; 2)
is symmetric; 3)
is
symmetric.
The pushforward of
g (20) results in the local form
g =
g
=
g
ij
(x, p)e
i
e
j
+
g
ab
(x, p)
e
a
e
b
,
where
g
ij
f =
i
k
j
l
g
kl
.
If is a linear connection on T
M as
X
Y f
(
X
Y )
for Y = f
(Y ) and X = f
) structure.
Proof. Locally, we can prove that the coecients of
connection
D
are related to coecients of
connection
D by formulas
L
k
ij
=
i
i
j
j
k
k
L
i
k
+
k
i
e
j
e
i
(x
i
),
C
i j
k
=
i
i
j
j
C
i
+
i
i
w
e
j
e
i
(x
i
).
Using formulas
D(
) = 0
D(
) = 0,
D
g =
D[f
(
g)] = f
(
D
g),
D
=
D[f
(
)] = f
(
D
),
where for
connections we use the symbol
D instead of , and by local
computations, we can verify that there are satised the conditions
D
g = 0
D
g = 0,
D
= 0
D
= 0.
24
D with coe
cients (42) generated by a Hamilton fundamental function H(x, p) which
is almost symplectic and metric compatible and adapted to symplectic mor
phism.
We can use
D(a )
_
e
(a) u
z
e
(a)
_
(
e
) +a d, (43)
where the local basis
z
e
is
e
1
on
W , are dened
(a) =
e
z
e
(a), and
1
(a) =
_
i
p+q
z
e
(a), if p +q > 0,
0, if p = q = 0,
(44)
where any a
W is homogeneous w.r.t. the grading deg
s
and deg
a
with deg
s
(a) = p and deg
a
(a) = q.
We note that the formulas (44) are dierent from (38) because they are
dened for dierent nonholonomic distributions and related adapted frame
structures.
Using dierential calculus of forms on
W , we prove
Proposition 4.3 The torsion and curvature canonical doperators of
T
z
(u)
e
, (45)
and
z
R
z
(u)
e
, (46)
where the nontrivial coecients of
and
D =
(
L
i
jk
,
C
k a
j
) (42) with respect to (40) and (41).
25
By straightforward verications, one gets the proof of
Theorem 4.4 Any FedosovHamilton operatorpairs (44) and related ex
tended operatorpair (43) are dened by torsion (45) and curvature (46)
following formulas:
_
D,
_
=
i
v
ad
Wick
(
z
T ) and
D
2
=
i
v
ad
Wick
(
z
R), (47)
where [, ] is the deg
a
graded commutator of endomorphisms of
W
and ad
Wick
is dened via the deg
a
graded commutator in (
W , ) .
The formulas (47) can be restated for any metric compatible dconnecti
on and
connection structures on T
D (42).
Theorem 5.1 A Hamilton fundamental function H(x, p) denes a at Fe
dosov
connection
D
+
D
i
v
ad
Wick
(
r)
satisfying the condition
D
2
= 0, where the unique element
r
W,
deg
a
(
r) = 1,
1
r = 0, is a solution of equation
r =
T +
R+
D
r
i
v
r
r.
26
The solution for
r can be computed recursively with respect to the total
degree Deg in the form
r
(0)
=
r
(1)
= 0,
r
(2)
=
T ,
r
(3)
=
1
_
R+
D
r
(2)
i
v
r
(2)
r
(2)
_
,
r
(k+3)
=
1
_
D
r
(k+2)
i
v
k
l=0
r
(l+2)
r
(l+2)
_
, k 1,
where a
(k)
is the Deghomogeneous component of degree k of an element
a
W .
Proof. It follows from a local component calculus with
adapted
coecients of
(T
M)[[v]] by formula
1
f
2
f ((
1
f)) ((
2
f)),
where the projection :
W C
(T
(T
M)[[v]]
W can be
calculated recursively w.r..t the total degree Deg,
(f)
(0)
= f and, for k 0,
(f)
(k+1)
=
1
_
D(f)
(k)
i
v
k
l=0
ad
Wick
(
r
(l+2)
)((f)
(kl)
)
_
.
Let
f
X the Hamiltonian vector eld corresponding to a function f
C
(T
M) on space (T
M,
) and consider the antisymmetric part
C(
1
f,
2
f)
1
2
_
C(
1
f,
2
f) C(
2
f,
1
f)
_
of bilinear operator C(
1
f,
2
f). A starproduct (34) is normalized if
1
C(
1
f,
2
f) =
i
2
{
1
f,
2
f}, where {, } is the Poisson bracket, see (13). For the
normalized , the bilinear operator
2
C denes a de RhamChevalley 2
cocycle, when there is a unique closed 2form
such that
2
C(
1
f,
2
f) =
1
2
(
f
1
X,
f
2
X) (48)
27
for all
1
f,
2
f C
(T
=
i
8
i
6
d
_
J
_
.
Let us dene the canonical class
, for the splitting (39). We can per
form a distinguished complexication of such second order tangent bundles
in the form T
C
_
TT
M
_
= T
C
(hT
M) T
C
(wT
M) and introduce
C
_
TT
M
_
= T
C
(hT
M)
T
C
(wT
D to T
C
_
TT
M
_
, we compute
the closed ChernWeyl form
= iTr
_
(h, v)
R(h, v)
T
_
(49)
= iTr
_
(h, v)
R
_
=
1
4
.
Using the canonical class
[
], we prove:
Theorem 5.3 The zerodegree cohomology coecient c
0
() for the almost
K ahler model of a Hamilton space
K
2n
, dened by a Hamilton fundamental
function H(x, p) is computed c
0
() = (1/2i)
.
The coecient c
0
() contains as a particular case the class of zerodegree
cohomologies computed for a metric of type g
i
j
(x) on M, dening a solution
of the Einstein equations and lifted on cotangent bundle by formula (1). In
such cases, this zerodegree coecient denes certain quantum properties of
the gravitational eld. A more rich geometric structure should be considered
if we dene a value similar to c
0
() encoding the information about Einstein
manifolds lifted to the cotangent bundle.
28
5.2 Quantum gravitational eld equations
Any solution in classical Einstein gravity can be embedded into a Hamilton
space model and quantized on cotangent bundle following the Fedosov quan
tization adapted to
and dconnections. Considering a dequantization
formalism [36], we construct certain quantum deformations of the classical
Einstein congurations in the classical limit. Such a model denes a non
holonomic almost K ahler generalization of the Einstein gravity on cotangent
bundle. The solutions for the cotangent gravity are, in general, with vio
lation of Lorentz symmetry induced by quantum corrections. The nature of
such quantum gravity corrections is dierent from those dened by Finsler
Lagrange models on tangent bundle (see, for instance, [33, 34, 43, 44, 45])),
locally anisotropic string gravity [40, 41, 42] with corrections from extra
dimensions and nonholonomic spinor gravity [37, 38, 39] and noncommu
tative gravity, see reviews of results in [29, 30]. The aim of this section
is to analyze how a generalization of Einstein gravity can be performed
on cotangent bundles in terms of canonical
connections, with geometric
structures induced by an eective Hamiltonian fundamental function, when
the Fedosov quantization can be naturally performed.
For a canonical
connections
D= {
} (42), we can dene the Ricci
tensor,
Ric(
D) = {
},
and the scalar curvature,
R
g
. On cotangent bundle T
M, we
postulate the eld equations
1
2
(
R +)
e
= 8G
, (50)
where
/u
is the
adapted base (40).
We consider the eective source 3form
du
du
du
,
where
_
e
+
e
_
= 8G
(51)
following from action by varying the components of e
, when
=
m
and
m
=
m
du
du
du
,
where
m
=
matter
S/e
.
Proof. It is a usual textbook and/or dierential form calculus (see, for
instance, [46, 14]). In our case, we have to use the
adapted bases (40)
and (41) for
D.
The ChernWeyl 2form (49) can be used to dene the quantum version
3
of Einstein equations (51) in the approaches with deformation quantization:
Corollary 5.1 The quantum eld equations on cotangent bundle generaliz
ing the Einstein gravitations in general relativity are
2G
J
. (52)
Proof. Multiplying
e
, we get the
almost symplectic form of the Einstein equations (52).
An explicit computation of
for nontrivial matter elds has to be per
formed for a deformation quantization model with interacting gravitational
and matter elds geometrized in terms of an almost K ahler model dened
for spinor and ber bundles on cotangent bundles.
6 Conclusions
In this paper we outlined a method of converting any regular Lagrange and
Hamilton dynamics into equivalent almost K ahler geometries with canon
ical nonlinear connection (Nconnection) and adapted almost symplectic
structures. The formalism was performed to be invariant under symplectic
morphisms and adapted to Legendre transforms of Lagrangians into Hamil
tonians and inversely. The geometry of cotangent bundles endowed with
3
in the sense of deformation quantization (i.e. when quantum equations are derived
following a deformation procedure) but not of perturbative quantum theory
30
nonholonomic distributions, formulated as certain HamiltonCartan spaces
being dual to the corresponding LagrangeFinsler spaces, described in Sec
tions 2 and 3 presents a key prerequisite of this approach to deformation
quantization.
Given a regular Lagrange (in particular, Finsler; and related Legendre
transforms), or Hamilton (in particular, Cartan), generating function, we
completely dene the fundamental geometric objects of a Hamilton geometry
modelled on cotangent bundle, inducing canonical almost symplectic and
compatible symplectic connection. The connectionpair and canonical
Viel mehr als nur Dokumente.
Entdecken, was Scribd alles zu bieten hat, inklusive Bücher und Hörbücher von großen Verlagen.
Jederzeit kündbar.