Professional Documents
Culture Documents
r
X
i
v
:
s
u
b
m
i
t
/
0
7
2
4
1
8
2
[
m
a
t
h
-
p
h
]
2
5
M
a
y
2
0
1
3
Almost Khler Ricci Flows and Einstein and
LagrangeFinsler Structures on Lie Algebroids
Sergiu I. Vacaru
Rectors Oce, Alexandru Ioan Cuza University, Alexandru Lapuneanu
street, nr. 14, UAIC Corpus R, oce 323; Iai, Romania, 700057
May 22, 2013
Abstract
In this work we investigate Ricci ows of almost Khler structures
on Lie algebroids when the fundamental geometric objects are com-
pletely determined by (semi) Riemannian metrics, or (eective) reg-
ular generating Lagrange/ Finsler, functions. There are constructed
canonical almost symplectic connections for which the geometric ows
can be represented as gradient ones and characterized by nonholonomic
deformations of Grigory Perelmans functionals. The rst goal of this
paper is to dene such thermodynamical type values and derive almost
Khler Ricci geometric evolution equations. The second goal is to
study how xed Lie algebroid, i.e. Ricci soliton, congurations can
be constructed for Riemannian manifolds and/or (co) tangent bun-
dles endowed with nonholonomic distributions modelling (generalized)
Einstein or Finsler Cartan spaces. Finally, there are provided some
examples of generic odiagonal solutions for Lie algebroid type Ricci
solitons and (eective) Einstein and LagrangeFinsler algebroids.
Keywords: Ricci ows, almost Khler structures, Lie algebroids,
Lagrange mechanics, Finsler geometry, eective Einstein spaces.
MSC: 53C44, 53D15, 37J60, 53D17, 70G45, 70S05, 83D99, 53B40, 53B35
Contents
1 Introduction: Almost Khler Models for Einstein & Lagran-
geFinsler Spaces 2
1.1 Preliminaries: nonholonomic manifolds and bundles . . . . . . 3
1.2 Canonical almost Khler variables for semiRiemannian and
LagrangeFinsler spaces . . . . . . . . . . . . . . . . . . . . . 6
2
L
y
a
y
b
, det [
h
ab
[ , = 0.
If V = TM, and n = m, we can consider / = L(x, y) as a regular La-
grangian dening a Lagrange space [23]. For L = F
2
(x, y), where F(x, y) =
F(x, y), > 0, we get a homogeneous Finsler generating function (addi-
tional conditions are imposed on F for dierent models of Finsler like ge-
ometries), see section 4.4. We shall work with arbitrary / considering the
Finsler congurations as certain particular ones distinguished by homogene-
ity conditions and additional assumptions.
Theorem 1.1 The EulerLagrange equations
d
d
L
y
i
L
x
i
= 0, where y
i
=
dx
i
/d, for x
i
() depending on real parameter , are equivalent to the "non-
linear" geodesic equations
dx
a
d
+2
G
a
(x, y) = 0, i.e. to the paths of a canonical
semispray S = y
i L
x
i
2
G
a
y
a
, when
G
a
=
1
4
h
a n+i
(
2
L
y
n+i
x
k
y
n+k
L
x
i
),
where
h
ab
is inverse to
h
ab
, and the set of coecients
N
a
i
=
G
a
y
n+i
denes a
Nconnection structure (1).
An explicit proof consists from straightforward computations. We put
"tilde" on some geometric objects in order to emphasize that they are gen-
erated by /.
Corollary 1.1 Any prescribed generating function / on V denes a canon-
ical distinguished metric, dmetric g, which is adapted to the Nconnection
2
Notational Remarks: All constructions in this work can be performed in coordi-
nate free form. Nevertheless, to nd/generate explicit examples of solutions of systems
of partial dierential equations, PDEs, is necessary to consider some adapted frame and
coordinate systems. Working with dierent classes of geometries and spaces, it is more
convenient to treat indices also as abstract labels. This simplies formulas and sugests
ideas how certain proofs are possible when local constructions have an "obvious" global
extension. For instance, index type formulas are largely used in G. Perelmans works and
for applications in mathematical relativity and/or geometric mechanics.
For a conventional n + m splitting, the local coordinates u = (x, y) can be labelled in
the form u
= (x
i
, y
a
), where i, j, k, ... = 1, 2, ...n and a, b, c... = n + 1, n + 2, ..., n + m,
where x
i
and y
a
are respectively the h and vcoordinates. A general local base is writ-
ten e
= (e
i
, e
a
) for some frame transforms e
= e
(u), where = /u
=
(i = /x
i
, a = /y
a
), and dene corresponding dual transforms with inverse matri-
ces e
(u), when e
= e
(u)du
= (e
i
=
i
N
a
i
a
, e
a
=
a
), e
= (e
i
= dx
i
, e
a
= dy
a
+
N
a
i
dx
i
)(3)
The proof is a similar to that for the Sasaki lift from M to TM, see [24].
It should be emphasized here that any metric g = g
on V, or TM,
can be expressed in a form g
= [ g
ij
,
h
ab
] (2), i.e. as a dmetric, via
corresponding frame transforms, e
= e
and g
= g
.
3
For
simplicity, we shall write (g, N) instead of
_
g,
N
_
if that will not result in
ambiguities. It is possible always to chose a necessary type / and encode
equivalently the geometric data with respect to a basis generated by
N[/],
or other conventional hvsplitting.
Denition 1.2 A distinguished connection (dconnection) D = (hD; vD)
is a linear connection preserving under parallelism the Whitney (1).
A dconnection is metric compatible if Dg = 0. We can also consider
hvsplitting (Nadapted decompositions) of tensors, vectors, dierential
forms, called respectively distinguished tensors, dtensors, distinguished vec-
tors, dvectors, etc. For instance, a dvector Y = hY +vY.
Denition 1.3 The torsion and curvature dtensors of a metric compatible
dconnection D are dened correspondingly
T (X, Y) := D
X
YD
Y
X+ [X, Y] and (4)
1(X, Y) := D
X
D
Y
D
Y
D
X
D
[X,Y]
. (5)
The Ricci, Ric, and Einstein, E, dtensors of D are constructed in stan-
dard form as for any linear connection, see relevant details in Appendix.
4
3
Prescribing a nonholonomic distribution L, we compute g
] = ee
e = w
= (L
i
jk
,
v
L
a
bk
; C
i
jc
,
v
C
a
bc
)} can be
derived for a 1form
:=
. The values
= {T
}, and dcurvature,
R
= {R
:=De
= de
= T
, R
:=D
= d
= R
,
see details in [17]. A Ricci dtensor Ric = {R
:= R
; R
= R
} splits into h
vcomponents R
= {Rij := R
k
ijk
, Ria := R
k
ika
, Rai := R
b
aib
, R
ab
:= R
c
abc
}. By
denition, the scalar curvature of a dconnection Dis
s
R := g
= R+S = g
ij
Rij +
h
ab
R
ab
, with R = g
ij
Rij and S = h
ab
R
ab
. The Einstein dtensor is E := Ric
1
2
g
s
R.
5
Theorem 1.2 For any given g and N, or g g and N
N (such values
are equivalent up to frame transforms if a generated function / is xed on
N), on a nonholonomic manifold V, we can construct three metric com-
patible linear connections determined completely by the metric eld. Such
connections are dened respectively by the conditions:
g = g
: g = 0;
T
D :
Dg = 0; h
= 0, v
D :
Dg = 0; h
= 0, v
D = +
Z and
D = +
Z, (6)
where all connections and distortion dtensors
Z and
Z are completely de-
ned by g for a prescribed N.
A proof follows form the computation of coecients of such values with
respect to Nadapted bases (3). The coecients of distortion dtensors
Z
and
Z are certain algebraic combinations of the coecients of, respectively,
torsion
T and
T (see details in [17]).
1.2 Canonical almost Khler variables for semiRiemannian
and LagrangeFinsler spaces
For any / and induced
N, we can dene a canonical almost complex
structure
J when
J(e
i
) = e
2+i
and
J(e
2+i
) = e
i
, for
J
J = I, for unity
matrix I.
5
The Neijenhuis tensor eld can be computed for any almost com-
plex structure J,
J
(X, Y) := [X, Y] + [JX, JY] J[JX, Y] J[X, JY].
On TV, any X = X
= X
i
e
i
+X
a
e
a
is a Nadapted dvector.
5
In component form, we compute
J =
J
e e
=
J
du
=
J
= e2+i e
i
+ei e
2+i
=
y
i
dx
i
+
x
i
N
2+j
i
y
j
dy
i
+
N
2+i
k
dx
k
.
6
Denition 1.4 An almost symplectic structure is dened by a nondegener-
ate 2form =
1
2
(u)e
.
In Nadapted form, =
1
2
ij
(u)e
i
e
j
+
1
2
ab
(u)e
a
e
b
.
Proposition 1.1 -Denitions: Any data (g, J) dene an almost symplec-
tic structure, (X, Y) := g (JX, Y) , for any dvectors X and Y.
An almost Hermitian model of a nonholonomic Riemannian manifold
(V, g, N) is dened by a triple H
n+n
= (V, (, ) := g (J, ) , J).
A space H
n+n
is almost Khler, denoted K
n+n
, if and only if d = 0.
Proofs consist from straightforward verications.
Lemma 1.1 Prescribing a generating function /(x, y) on V, we can model
equivalently this nonholonomic manifold as a canonical almost Khler space.
Proof. For any g = g, N =
N and J =
J canonically determined by /,
we can dene
(, ) := g
_
J,
_
. In Nadapted form,
=
1
2
(u)du
du
=
1
2
(u)e
(7)
= g
ij
(x, y)e
n+i
dx
j
= g
ij
(x, y)(dy
n+i
+
N
n+i
k
dx
k
) dx
j
,
where
ab
=
n+i n+j
are respectively the coecients
ij
. We obtain d
=
dd = 0 if
= d , for :=
1
2
L
y
i
dx
i
. (end proof).
Due to M. Matsumoto [25] (the original results were proven for Finsler
Cartan spaces and then extended to Lagrange geometry and nonholonomic
manifolds, see references in [20]) we can prove this theorem.
Theorem 1.3 The Cartan dconnection
D is a unique almost symplectic
dconnection which satises the conditions
D
= 0 and
D
J = 0.
Let us denote by Ric,
Ric and
Ric the Ricci tensors for, respectively,
linear connections ,
D and
D all dened by the same metric g and corre-
sponding N, or /.
Theorem 1.4 An Einstein manifold is dened by a metric g as a solution
of the vacuum gravitational eld equations with cosmological constant ,
Ric = g; (8)
or of equivalent equations in nonholonomic canonical variables,
(M).
If N is integrable (i.e. with trivial Nconnection structure), a Lie d
algebroid is just a Lie algebroid c = (E, , , ). For "non-boldface" con-
structions, see details in Refs. [28, 29, 30, 31, 32, 33]. The anchor map is
equivalent to a homomorphism between the Lie algebras (Sec(), , ) and
(A(M), , ) .
6
6
Locally, the properties of a Lie dalgebroid E are determined by the functions
i
a
(x
k
)
and C
e
ab
(x
k
), where x = {x
k
} are local coordinates on a chart U on M (or on hV if a non-
9
Example 2.1 Nonholonomic Lie algebroids: If E =TV for a nonolo-
nomic manifold V = (V, N) with Nsplitting (1), the values like A(M) are
considered for M hV and sections are modelled on vV. We shall con-
struct Ricci soliton congurations, or Einstein manifolds, with Lie algebroid
symmetry in section 4.
Let us extend the concept of prolongation Lie algebroid [30, 31, 32, 33] (in
our case) in order to include Nconnections. We consider a Lie dalgebroid
c = (E, , , ) and a bration : P M both dened over the same
manifold M. In general, E, P and TM, or TV, may be enabled with dierent
Nconnection structures. The local coordinates are u
= (x
i
, y
A
) P when
e
a
will be used for a local basis of sections of E. If E = TV, we write
u
= (x
i
, y
I
). In our constructions, we can consider that P = E. The anchor
map : E TM and the tangent map T : TP TM are all dened to
be hvadapted for nonholonomic bundles and/or bered structures. These
structures can be used to construct the subset
T
E
s
P := (b, v) E
x
T
x
P; (b) = T
p
(v); p P
x
, (p) = x M (13)
and prove this result:
Theorem 2.1 Denition. The the prolongation T
E
P :=
sS
T
E
s
P of
a nonholonomic E over is another Lie dalgebroid (the construction (13)
can be considered for any set of charts covering such spaces).
Similarly to holonomic congurations, with trivial Nconnection struc-
ture, the prolongation Lie dalgebroid T
E
P is called the (nonholonomic)
Etangent bundle to , which is also a nonholonomic vector bundle over P.
The corresponding projection
E
P
is just onto the rst factor,
E
P
(b, v) = b
being adapted to the Nconnection structures [17]. The elements of T
E
P are
parameterized by Nadapted triples (p, b, v). We shall use also brief denota-
tions (p, b, v) T
E
P (b, v) T
E
P if that will not result in ambiguities.
The Nadaped anchor
: T
E
P T P is given by maps
(p, b, v) = v,
holnomic manifold is considered), with (ea) =
i
a
(x)ei and ea, e
b
= C
f
ab
(x)e
f
, satisfying
the following equations
i
a
ei
j
b
i
b
ei
j
a
=
j
f
C
f
ab
and
cycl (a,b,f)
i
a
iC
f
be
+ C
d
be
C
f
ad
= 0.
Boldface operators are dened by Ncoecients similarly to (3). For trivial Nelongated
partial derivatives and dierentials, we can use local coordinate frames when ei i,
ea = a, e
i
= dx
i
, e
a
dy
a
.
The exterior dierential on E with nonholonomic E can be dened in standard form
with E. We introduce on E the operator d : Sec(
) Sec(
k+1
), d
2
= 0, where
) Sec(
), using the
cohomology operator d and its inverse iX.
10
i.e. projection onto the third factor when the hcomponents transforms
in other hcomponents etc. For more special cases, we can dene also the
projection onto the second factor (i.e. a morphism of Lie dalgebroids over
), T : T
E
P E, when T (p, b, v) = b. An element (p
1
, b
1
, v
1
) T
E
P
is vertical if T (p
1
, b
1
, v
1
) = b
1
= 0, i.e. such elements are of type (p, 0, v)
when v is a vertical vector (tangent to P at point p).
To understand consequences of Theorem 2.1 let us consider some local
constructions. In coecient form, any element of a prolongation Lie d
algebroid T
E
P can be parameterized z = (p, b, v) T
E
P, for b = z
a
e
a
and
v =
i
a
z
a
e
i
+ v
A
A
, for /y
A
, can be decomposed z = z
a
A
a
+ v
A
1
A
. The
couple (A
a
, 1
A
) , with vertical 1
A
, denes a local basis of sections of T
E
P. For
such bases, we can write A
a
= A
a
(p) =
_
e
a
((p)),
i
a
e
i|p
_
and 1
A
=
_
0,
A|p
_
,
where partial derivatives and their Nelongations are taken in a point p S
x
.
It is also possible to elaborate on prolongation Lie dalgebroids a Nadapted
exterior dierential calculus. The anchor map
(Z) =
i
a
Z
a
e
i
+V
A
A
is an
Nelongated operator acting on sections Z with associated decompositions
of type z. The corresponding Lie brackets are
A
a
, A
b
= C
f
ab
A
f
, A
a
, 1
B
= 0, 1
A
, 1
B
= 0.
Denoting by
_
A
a
, 1
B
_
the dual bases to (A
a
, 1
A
) , we can elaborate a dier-
ential calculus for Nadapted dierential forms using
dx
i
=
i
a
A
a
, for dA
f
=
1
2
C
f
ab
A
a
A
b
, and dy
A
= 1
A
, for d1
A
= 0.
(14)
Nconnections can be introduced on T
E
P similarly to (1).
Denition 2.2 On a prolongation Lie dalgebroid, a Nconnection is de-
ned by a hvsplitting,
A : T
E
P = hT
E
PvT
E
P. (15)
We can consider A : T
E
P T
E
P, with A
2
= id, as a nonholonomic
vector bundle, and Lie dalgebroid, morphism dening an almost product
structure on
P
: TP P, for a smooth map on TP0, were 0 denotes
the set of null sections. Any Nconnection A induces h- and vprojectors
for every element z = (p, b, v) T
E
P, when h(z) =
h
z and v(z) =
v
z, for
h =
1
2
(id + A) and v =
1
2
(id A). The respective h- and vsubspaces are
hT
E
P = ker(id A) and vT
E
P = ker(id +A).
Let us analyze some local constructions related to Aconnection struc-
tures for prolongation Lie dalgebroids. Locally, Nconnections are deter-
mined respectively by their coecients N = N
A
i
and A= A
A
a
, when
N =N
A
i
(x
k
, y
B
)dx
i
A
and A = A
A
a
A
a
1
A
. (16)
Such structures on TP and T
E
P are compatible if A
A
a
= N
A
i
i
a
. Using A
A
a
,
we can generate sections
a
:= A
a
A
A
a
1
A
as a local basis of hT
E
P.
11
Corollary 2.1 Any Nconnection A
A
a
on T
E
P determines a Nadapted
frame structure
e
:=
a
= A
a
A
C
a
1
C
, 1
A
, (17)
and its dual
e
:= A
a
,
B
= 1
B
+A
B
c
A
c
. (18)
Proof. In above formulas, the overlined small Greek indices split =
(a, A) if an arbitrary vector bundles P is considered, or = (a, b) for P = E.
A proof follows from an explicit construction such Nadapted frames. Then
the formulas can be considered for arbitrary frames of references.
The Nadapted bases (17) satisfy the relations e
= W
,
with nontrivial anholonomy coecients W
= C
f
ab
,
C
ab
,
B
A
C
c
. The cor-
responding generalized Lie brackets are
a
,
b
= C
f
ab
f
+
C
ab
1
C
,
a
, 1
B
= (
B
A
C
a
)1
C
, 1
A
, 1
B
= 0. (19)
Denition 2.3 The curvature of Nconnection A
A
a
is by denition the Neig-
enhuis tensor
h
N of the operator h,
h
N(, ) = h, h
hh,
h, h
+h
2
h, h
=
1
2
C
ab
A
a
A
b
1
C
,
where
C
ab
=
b
A
C
a
a
A
C
b
+C
f
ab
A
C
f
. (20)
It should be noted that above formulas for T
E
P mimic (on sections of E
for P = E) the geometry of tangent bundles and/or nonholonomic manifolds
of even dimension, endowed with Nconnection structure. On applications
in modern classical and quantum gravity, with various modications, and
nonholonomic Ricci ow theory, see Refs. [18, 12, 15]. If P ,= E, we model
nonholonomic vector bundle and generalized Riemann geometries on sections
of T
E
E.
2.2 Canonical structures on Lie dalgebroids
Almost Khler Lie algebroid geometries can be modelled on prolonga-
tion of Lie dalgebroids following the method outlined in section 1.2 for
vector/tangent bundles and nonholonomic manifolds.
2.2.1 dconnections and dmetrics on T
E
P
Denition 2.4 A distinguished connection, dconnection, T = (hT, vT),
on T
E
P is a linear connection preserving under parallelism the splitting (15).
The formulas for distinguished torsion (4) and curature (5) can be gen-
eralized for prolongation Lie dalgebroids.
12
Denition 2.5 The torsion and curvature for any dconnection T on T
E
P
are dened respectively by
T (x, y) := T
x
y T
y
x +x, y
, and
1(x, y)z :=
_
T
x
T
y
T
y
T
x
T
x,y
_
z.
Let us consider sections x, y, z of T
E
P when, for instance z = z
=
z
a
a
+z
A
1
A
, or z = hz +vz. Using the rules of absolute dierentiation (14)
for Nadapted bases e
:=
a
, 1
A
and e
:= A
,
B
and associating to
T a dconnection 1form
:=
and curvature 1 = R
2forms,
see details in Appendix.
We can introduce a metric structure as a nondegenerate symmetric sec-
ond rank tensor g = g
.
Proposition 2.1 Denition. Any metric structure on T
E
P can be rep-
resented in Nadapted form as a dmetric g=hgvg adapted to the Whitney
sum (15).
Proof. It follows from an explicit construction with respect to N
adapted frames, see formula (A.3).
If a dmetric g and a dconnection T are independent geometric struc-
tures, such values are characterized (additionally to T and 1) by a non-
metricity eld Q(y) := T
y
g, with Nadapted coecients Q
= T
.
Proposition 2.2 Denition. The data (g, T) are metric compatible if
Q = Tg = 0 holds in Nadapted form for any h and vcomponents.
Proof. It follows from a straightforward computation when the co-
ecients of dmetric g
= y
a
a
+ y
A
1
A
. Such a condition splits into respective conditions
for h-vcomponents which with respect to Nadapted frames are T
f
g
ab
=
0, T
A
g
ab
= 0, T
f
g
AB
= 0, T
C
g
AB
= 0.
2.2.2 The canonical dconnection
On T
E
P, we can dene two important metric compatible linear connec-
tion structures g
T + v
T, when
the h-vsplitting is determined by (15).
Theorem 2.2 There is a canonical dconnection
T completely dened by
data (A, g
) for which
Tg = 0 and h- and v-torsions of
T are prescribed,
respectively, to be with Nadapted coecients
T
a
bf
= C
a
bf
and
T
A
BC
= 0.
13
Proof. See proof in section A.2.
For trivial algebroid structures,
T
D following the conditions of
Theorem 1.2.
Remark 2.1 There is a canonical distortion relation
T = +
:,
when both linear connections
T and (the last one is the LeviCivita
connection) and the distorting dtensor
: are dened by the same data
(A, g
). This is a generalization on T
E
P of the the formula (6). In
explicit form, the Nadapted coecients of such values are computed
following formulas (A.5) and (A.6).
We note that h
= 0 for
D on TMbut h
,= 0 for
T on T
E
P. The
formulas for
L
a
bf
in (A.5) contain additional terms with C
a
bf
which
results in nontrivial
T
a
bf
= C
a
bf
and additional terms in Nadapted
coecients R
a
ebf
and R
A
Bbf
of curvature (A.2).
In order to generate exact solutions, it may be more convenient to work
with an auxiliary dconnection
c
T := +
c
: for which h
c
= 0 and
v
c
T
a
bf
= 0 and
c
T
A
BC
= 0 but,
in general,
1 , =
c
1and
T ,=
c
L
a
bf
=
1
2
g
ae
(
f
g
be
+
b
g
fe
e
g
bf
) and
c
L
A
Bf
=
L
A
Bf
,
c
B
C
=
B
C
,
c
B
A
BC
=
B
A
BC
are those from (A.5).
2.3 Almost symplectic geometric data
2.3.1 Semispray congurations and Nconnections
We show how a canonical almost symplectic structure can be generated
on T
E
P by any regular eective Lagrange function /.
Lemma 2.1 Prescribing any (eective) Lagrangian /, we can construct a
canonical Nconnection
q
A := /i
q
S dened by a semispray q = y
a
A
a
+
q
A
1
A
and Lie derivative /i
q
acting on any X Sec(TE) following formula
q
A(X) = q, SX
+S q, X
.
Proof. We use the semispray formula Sq = with the operators S
and from (A.9) and compute
q
A(A
a
) = q, S(A
a
)
+S q, A
a
= A
a
+ (
a
q
b
+y
f
C
b
fa
)1
b
.
For
q
A(1
a
) = 1
a
and
q
A(A
a
) = A
a
2
q
A
f
a
(x, y)1
f
, we dene the
Nconnection coecients A
f
=
1
2
(
a
q
f
+y
b
C
f
ba
), see formulas (16).
We can formulate on Lie daglebroids the analog of Theorem 1.1.
14
Theorem 2.3 Any eective regular Lagrangian / C
A =
A
f
a
=
1
2
(
a
f
+y
b
C
f
ba
), (21)
determined by semispray congurations encoding the solutions of the Euler
Lagrange equations (A.12).
Proof. It is a straightforward consequence of above Lemma and (A.11).
To generate Nconnections, we can use sections
L
= y
a
A
a
+
a
1
a
, with
q
e
=
e
(x
i
, y
b
) (A.11). Let us consider the value S = y
a
i
a
L
x
i
2
G
A
y
A
,
where
G
A
is associated to the "nonlinear" geodesic equations for sections,
dy
A
d
+ 2
G
A
= 0, depending on real parameter , and dene a Nconnection
structure
N
F
a
=
G
F
y
a
of type (16). For P = E, such sections can be related
to the integral curves of the EulerLagrange equations (A.12) if we chose
the sections
G
F
G
f
and
f
(x
i
, y
b
) in such forms that
A
f
a
=
1
2
(
a
f
+
y
b
C
f
ba
) =
G
f
y
a
. The constructions can be performed on any chart covering
such spaces, i.e. we can prove that the coecients (21) dene a Nconnection
structure (15).
Proposition 2.3 Any metric structure on T
E
P can be represented in N
adapted form as a dmetric g=h gv g constructed as a formal Sasaki lift
determined by an eective regular generating function /.
Proof. Sasaki lifts are used for extending certain metric structures from
a base manifold, for instance, to the total space of a tangent bundle, see
details in [24]. For vector bundles, the formulas (2) provide a typical ex-
ample of such a construction using canonical Nconnection structure
N.
The method can be generalized for prolongation Lie dalgebroids. At the
rst step, we use the canonical Nconnection
A =
A
f
a
(21) and construct
Nadapted frames of type (17) and (18), respectively,
e
:=
a
= A
a
A
f
a
1
f
, 1
b
and e
:= A
a
,
b
= 1
b
+
A
b
f
A
f
. (22)
Then (the second step) we dene a canonical dmetric
g := g
= g
ab
A
a
A
b
+ g
ab
b
(23)
using the Hessian (A.10). Considering an arbitrary dmetric g = g
(A.3), we can nd a regular / and certain frame transforms e
= e
when g
= e
and g
.
15
2.3.2 RiemannLagrange almost symplectic structures
Let us consider canonical data dened respectively by Nelongated frames
e
= (e
a
=
a
, 1
A
) (22), Nconnection
A (21) and dmetric g = g (23).
Proposition 2.4 Denition. For any regular eective Lagrange structure
/, we can dene a canonical almost complex structure on T
E
E following
formulas
(e
a
) = 1
m+a
and
(1
m+a
) = e
a
, when
= I
Proof. It follows from an explicit construction of a dtensor eld
= 1
m+a
A
a
+e
a
a
.
Using vielbeins e
and e
on T
E
E, when
= e
.
In general, we can dene an almost complex structure on T
E
E for
an arbitrary Nconnection A (15) by using Nadapted bases (17) and (18)
which are not necessarily induced by an eective Lagrange function /. This
allows us to generate almost Hermitian models and not almost Khler ones.
Denition 2.6 The Nijenhuis tensor eld for any almost complex structure
on T
E
E determined by a Nconnection A (equivalently, the curvature of
Nconnection A) is by denition
J
(x, y) := [x, y]+[x, y][x, y][x, y], (24)
for any sections x, y of T
E
E.
We can introduce an arbitrary almost symplectic structure as a 2form
on a prolongation Lie dalgebroid.
Denition 2.7 An almost symplectic structure on T
E
P is dened by a non-
degenerate 2form =
1
2
(x
i
, y
B
)e
.
Using frame transforms, we can prove
Proposition 2.5 For any on T
E
P when h(x,y) := (hx,hy), v(x,y) :=
(vx,vy), there is a unique Nconnection A = A
A
a
(16) when
= (hx, vy) = 0 and = h +v. (25)
Proof. In Nadapted form,
=
1
2
ab
(x
i
, y
C
)A
a
A
b
+
1
2
AB
(x
i
, y
C
)
A
B
, (26)
where the rst term is for h and the second term is v, i.e. we get the
second formula in (25).
16
Denition 2.8 a) An almost Hermitian model of a prolongation Lie
dalgebroid T
E
E equipped with a Nconnection structure A is dened
by a triple 1
E
E = (T
E
E, , ), where (x,y):= g(x,y).
b) A Hermitian prolongation Lie dalgebroid 1
E
E is almost Khler,
denoted /
E
E, if and only if d = 0.
For eective regular Lagrange congurations, we can formulate:
Theorem 2.4 Having chosen a generating function /, we can model equiv-
alently a prolongation Lie dalgebroid T
E
E as an almost Khler geometry,
i.e. 1
E
E = /
E
E.
Proof. For the canonical geometric data (g = g,
A,
), we dene the
symplectic form
(x,y):= g(x,y) for any sections x, y of T
E
E. In local
Nadapted form,
= g
ab
a
A
b
. Let us consider the form :=
1
2
L
y
m+a
A
a
.
Using Proposition 2.5 and Nconnection
A (21), we prove that
= d ,
which means that d
= e
=
1
2
ab
(x
i
, y
C
)A
a
A
b
+
1
2
AB
(x
i
, y
C
)
B
. (27)
T = h
T = (
h
T
a
,
v
T
a
); v
T = (
h
T
A
,
v
T
A
)
=
= (
L
a
bf
,
L
A
Bf
;
B
a
bC
,
B
A
BC
),
we can verify that
T =
_
h
T = (
h
T
a
,
v
T
a
); v
T = (
h
T
A
,
v
T
A
)
_
=
= (
L
a
bf
,
L
A
Bf
;
B
a
bC
,
B
A
BC
), with
17
L
a
bf
=
L
a
bf
+
1
2
ae
h
T
f
be
,
L
A
Bf
=
L
A
Bf
+
1
2
AE
v
T
f
EB
,(28)
B
a
bC
=
B
a
bC
+
1
2
ae
h
T
C
be
,
B
A
BC
=
B
A
BC
+
1
2
AE
v
T
C
EB
,
satises the conditions
h
T
a
be
= 0,
v
T
a
AB
= 0,
h
T
A
be
= 0,
v
T
A
AB
= 0,
which are h- and vprojections of
T
x
= 0 from Denition 4.1.
Let us introduce the operators
ab
cd
=
1
2
(
a
c
b
d
cd
ab
) and
AB
CD
=
1
2
(
A
C
B
D
CD
AB
). (29)
Theorem 2.5 The set of dconnections
s
= (
s
L
a
bf
,
s
L
A
Bf
;
s
B
a
bC
,
s
B
A
BC
)
labeled by an abstract left index "s", compatible with an almost symplectic
structure (26), are parameterized by
s
L
a
bc
=
L
a
bc
+
da
be
Y
e
dc
,
s
L
A
Bc
=
L
A
Bc
+
CA
BE
Y
E
Cc
, (30)
s
B
a
bC
=
B
a
bC
+
ea
bf
Y
f
eC
,
s
B
A
BC
=
B
A
BC
+
EA
BF
Y
F
EC
,
where the Nadapted coecients are given by (28), the operators are those
from (29) and Y
= (Y
e
dc
, Y
E
Cc
, Y
f
eC
, Y
F
EC
) are arbitrary dtensor elds.
Proof. It follows from straightforward Nadapted computations.
Remark 2.2 The Lie algebroid structure functions C
d
bf
in (A.5) can be con-
sidered as an example of dtensor elds Y
in (30). On T
E
E, we can
work as on TM but for diernet classes of nonholonomic distributions for
sections. The dconnections
T,
T can be constructed for correspondingly
dened Nconnection structures A,
A when the main geometric properties
are similar to some geometric models with
D,
D and certain N,
N. The
nonholonomic frame structures on Lie dalgebroids are dierent from those
on nonholonomic tangent bundles because in the rst case the vierbein elds
encode the ancor and Lie type structure functions.
We can select a subclass of metric and/or almost symplectic compatible
dconnections on T
E
E which are completely dened by g and prescribed by
an eective Lagrange structure /(x, y).
Theorem 2.6 On T
E
E, there is a unique normal dconnection
n
T =
h
n
T = (
n
h
T
a
=
T
a
,
n
v
T
a
=
T
a
); v
n
T = (
n
h
T
A
=
T
A
,
n
v
T
A
=
T
A
)
=
n
= (
n
L
a
bf
=
L
a
bf
,
n
L
A=m+a
B=m+b f
=
L
a
bf
;
n
B
a=Am
b=Bm C
=
B
A
BC
,
n
B
A
BC
=
B
A
BC
), (31)
which is metric compatible,
T
a
g
bc
= 0 and
T
A
g
BC
= 0, and completely
dened by g = g and a xed /(x, y).
18
Proof. We provide a proof constructing such a normal dconnection in
explicit form an example when
n
T =
T generalizes the concept of Cartan
dconnection from Theorem 1.3. Such a dconnection is completely dened
by couples of h and vcomponents
T = (
T
a
,
T
A
), i.e.
= (
L
a
bf
,
B
A
BC
).
Let us chose
L
a
bf
=
1
2
g
ae
(
f
g
be
+
b
g
fe
e
g
bf
),
B
A
BC
=
1
2
g
AD
(1
C
g
BD
+1
B
g
CD
1
D
g
BC
),
(32)
where the Nelongated derivatives are taken in the form (22) and g
ab
=
g
A=m+a B=m+b
are generated by canonical values using the Hessian (A.10)
and (23) induced by a regular /(x, y), we can prove that this dconnection
is unique and satises the conditions of the theorem. Via frame transforms,
we can consider any metric structure g g.
Finally, we formulate this very important for our purposes result:
Theorem 2.7 The normal dconnection
n
T =
T denes a unique almost
symplectic dconnection,
T
T, see Denition 2.9, which is Nadapted
and compatible to the canonical almost symplecti form
(27), i.e.
T
=0
and
T
a
cb
=
T
A
CB
= 0, see torsion coecients (A.16).
Proof. Using the coecients (32), we can check that such a normal
dconnection satises the conditions of this theorem.
Conclusion 2.1 Prescribing an eective generating function / on T
E
E,
we can transform this prolongation Lie dalgebroid into a canonical almost
Khler one, /
E
E. It is possible to work equivalently with any geometric data
_
g, A,
T = +
:
_
_
g,/,
A,
T
_
(, ):= g(
,),
T
_
.
The Lie algebroid structure functions (
i
a
, C
f
ab
) are encoded into nonholo-
nomic distributions on T
E
E determining such equivalent prolongation Lie
dalgebroid congurations.
2.4 Almost Khler Einstein and Lagrange Lie dalgebroids
We can formulate analogs of Einstein equations for dierent classes of
dconnections on prolongation Lie dalgebroids and almost Khler models.
Corollary 2.2 Denition The Ricci tensor of a dconnection T on a
T
E
P endowed with dmetric structure g is dened following formula 1ic =
R
:= R
=
R
= R
ab
:= R
c
abc
, R
aA
:= R
c
acA
, R
Aa
:= R
B
AaB
, R
AB
:= R
C
ABC
.
(33)
19
Proof. The formulas for hvcomponents (33) are respective contrac-
tions of the coecients (A.2).
The scalar curvature
s
R of T is by denition
s
R := g
= g
ab
R
ab
+g
AB
R
AB
. (34)
Using (33) and (34), we compute the Einstein dtensor E
:= R
1
2
g
s
R of T. Such a tensor can be used for modeling eective gravity
theories on sections of T
E
P with nonholonomic frame structure [26, 17, 27].
The conditions of Theorem 1.3 can be reformulated and proven on T
E
E,
or /
E
E, respectively, for
T and
T =
T. Prescribing the geometric data
(/;
i
a
, C
f
ab
), any solution of the gravitational eld equations (9) and (10) can
be transformed into solutions of
(, ):= g(
,),
T = +
:
_
.
The functions determining nonholonomic distributions, Nconnection coe-
cients and Lie algebroid structure functions are considered to be of the same
smooth class as the coecients of g and .
The theory of LagrangeRicci ows on T
E
E is formulated for evolving
nonholonomic dynamical systems on the space of equivalent geometric data
_
/ : g,
_
, when the functionals
T and
J are postulated to be of Lyapunov
type, see below formulas (40) and (41). Ricci solitonic congurations are
dened as xed on points of the corresponding dynamical systems. The
"stationary" variational conditions depend on what type of the Ricci tensor
we use, for instance, that one for the connections or
T. We can elaborate
Nadapted almost Khler scenarios if the Perelmans functionals are re
dened in terms of geometric data ( g,
T) and the derived ow equations are
considered in Nadapted variables. Both approaches are equivalent if the
distortion relations are considered for the same family of metrics, g() = g()
correspondingly computed for a set /().
Lemma 3.1 For the scalar curvature and Ricci tensor determined by the
distortion relation
T =
T = +
:, (37)
the Perelmans functionals are dened equivalently in almost Khler canoni-
cal Nadapted variables,
T( g,
T,
f) =
_
V
(
s
R+[h
f[
2
+[ v
T
f[)
2
)e
f
dv, (38)
J( g,
T,
f, ) =
_
V
[ (
s
R+[h
f[ + [v
T
f[)
2
+
f 2m] dv, (39)
where the new scaling function
f is intorduced for
_
V
dv = 1 with volume
element dv, = (4 )
m
e
f
and > 0, where =
h
=
v
for a couple
of possible h and vows parameters, = (
h
,
v
).
Proof. On T
E
P, evolution models can be formulated as in standard
theory for Riemann metrics [1, 2, 4, 5, 6] but for a family of geometric
data
_
g(), ()
_
induced by a family of regular /() C
(P) with a ow
parameter [, ] R, when > 0 is taken suciently small. For P = E,
we can postulate on the space of Sec(E), for : E M, dimE = n+m 3
and dimM = n 2, the (Perelmans) functionals
T(g, , f, ) =
_
V
_
R +
2
_
e
f
dv, (40)
J(g, , f, ) =
_
V
_
_
R +
_
2
+f 2m)
_
dv, (41)
22
where the volume form dv and scalar curvature
R of are computed for sets
odiagonal metrics g
2
)e
f
=
(
s
R+ [
f[
2
)e
f
+
. We obtain the Nadapted functional (38). For the
second functional (41), we rescale and write
_
(
R +
)
2
+f 2m
= [ (
s
R+[h
f[ +[ v
T
f[)
2
+
f 2m] +
1
,
for some
and
1
for which
_
V
dv = 0 and
_
V
1
dv = 0. Finally, we get
the formula (39).
In this section, we shall only sketch the key points for proofs of theorems
when the geometric constructions are straightforward consequences of those
presented for the LeviCivita connection in [4, 5, 6, 7, 8, 9] and extended to
nonholonomic congurations in [12, 13, 14, 15, 16, 17]. For our models, we
consider operators which up to frame transforms are dened by / via on
T
E
E. Following Remark 3.1 for distortions completely dened by g, we can
study canonical Cartan and almost Khler Ricci ows on prolongation Lie
algebroids as nonholonomic deformations of the Riemannian evolution.
Using
(,):= g( ,), we can dene the canonical (almost symplectic)
Laplacian operator,
:=
T
1
dv = 0 if those values
can be generated, or transformed into gradient type ones, via nonholonomic
deformations. Inverting the distortion relations (37), we can compute
=
+
Z
,
Z
=
:
(
:
) +
:
]; (42)
R
=
R
+
Zic
,
R =
s
R+ g
Zic
=
s
R+
s
Z,
s
Z = g
Zic
= h
Z +v
Z, h
Z = g
ab
Zic
ab
, v
Z = g
AB
Zic
AB
;
s
R = hR +vR, hR := g
ab
R
ab
, vR = g
AB
R
AB
,
where the terms with left up label "Z" are determined by
: (for instance,
Zic
+R
) and R
[]
:=
1
2
(R
). Using
23
deformations of type (42) of corresponding geometric values in the proof, for
instance, given in for Proposition 1.5.3 of [7], we obtain
Theorem 3.1 a) The canonical Nadapted Ricci ows for the Cartan d
connection
T preserving a symmetric metric structure g on T
E
E can be
characterized by this system of geometric ow equations:
g
ab
= (
R
ab
+
Zic
ab
),
g
AB
= (
R
AB
+
Zic
AB
), (43)
R
aA
=
Zic
aA
,
R
Aa
=
Zic
Aa
, (44)
= (
+
Z
)
f +
T
:
_
f
R
s
Z, (45)
and the property that
T( g,
T,
f) =
_
V
[[
R
ab
+
Zic
ab
+ (
T
a
:
a
)(
T
b
:
b
)
f[
2
+
[
R
AB
+
Zic
AB
+ (
T
A
:
A
)(
T
B
:
B
)
f[
2
]e
f
dv,
_
V
e
f
dv = const.
b) In almost symplectic variables with
= g
ab
a
X
b
and almost Khler
dalgebroids /
E
E, and for redened scaling function
f, up to normalizing
terms, the h- and vevolution equations are written in equivalent form
ab
R
[ab]
,
AB
R
[AB]
. (46)
For dierent classes of distortions of type (42), we can redene the scaling
functions from above Lemma and write the evolution equations (43) in the
form (46) for symplectic variables with (27). On T
E
E, the corresponding
system of Ricci ow evolution equations can be written for
T,
g
ab
= 2
R
ab
,
g
AB
= 2
R
AB
, (47)
R
aA
= 0,
R
Aa
= 0,
f
f +
R,
which can be derived from the functional
T( g,
T,
f) =
_
V
(
s
R + [
f[
2
)
e
f
dv. We note that the conditions of type
R
A
= 0 and
R
A
= 0 must be
imposed in order to model Nadapted evolution scenarios only with symmet-
ric metrics. In general, a nonholonomically constrained evolution can result
in nonsymmetric metrics (see examples in [36]).
Corollary 3.1 The geometric almost Khler dalgebroid evolution dened
in Theorem 3.1 is characterized by corresponding ows (for all time
24
[0,
0
)) of Nadapted frames, e
() = e
(, x
i
, y
C
)
, which up to frame/
coordinate transforms are dened by the coecients
e
(, x
i
, y
C
) =
_
e
a
a
(, x
i
, y
C
)
N
B
b
(, x
i
, y
C
) e
a
B
(, x
i
, y
C
)
0 e
A
A
(, x
i
, y
C
)
_
,
e
(, x
i
, y
C
) =
_
e
a
a
=
a
a
e
b
i
=
N
B
b
(, x
i
, y
C
)
b
a
e
a
A
= 0 e
A
A
=
A
A
_
,
with g
ab
() = e
a
a
(, x
i
, y
C
) e
b
b
(, x
i
, y
C
)
a
b
and
g
AB
() = e
A
A
(, x
i
, y
C
) e
B
B
(, x
i
, y
C
)
A
B
, where
a
b
= diag[1, ..., 1] and
B
= diag[1, ..., 1] in order to x a Riemannian signature of g
[0]
(x
i
, y
C
),
is given by equations
= g
= (
+
Z
)
f +
T
a
:
a
)
f
R +
2m
,
= 1,
can be derived for a functional
J satisfying the condition
J( g(),
f(), ()) = 2
_
V
[[
Zic
+ (
)(
)
f
1
2
g
[
2
](4 )
m
e
f
dv,
for
_
V
e
f
dv = const. Such a functional is Nadapted and nondecreasing if
it is both h and vnondecreasing.
25
Proof. For the LeviCivita connection on T
E
E, the proof is similar to
that in Proposition 1.5.8 in [7] containing the details of the original result
from [4]. Using Nadapted deformations, the geometric constructions are
performed in almost symplectic variables on /
E
E.
Let us remember some main concepts from statistical thermodynamics.
It is considered a partition function Z =
_
exp(E)d(E) for a canonical
ansamble at temperature
1
. Such a temperature is dened by the mea-
sure determined by the density of states (E). We can provide a statistical
analogy computing respective thermodynamical values. In standard form,
there are introduced E := log Z/, the entropy S := E + log Z
and the uctuation :=
_
(E E)
2
_
=
2
log Z/
2
. The original idea of
G. Perelman was to use such values for characterizing Ricci ows of Rieman-
nian metrics [4]. The constructions can be elaborated in Nadapted form
for geometric ows subjected to nonintegrable constraints on various spaces
endowed with nonholonomic distributions of commutative and noncommu-
tative type, Lie algebroids etc [12, 13, 14, 15, 16, 17].
Theorem 3.3 The Nadapted metric compatible (with symmetric metrics)
Ricci on T
E
E are characterized by a) canonical thermodynamic values
_
E
_
=
2
_
V
(
s
R+[
f[
2
m
) dv,
S =
_
V
[ (
s
R+[
f[
2
) +
f 2m] dv,
= 2
4
_
V
[[
Zic
+ (
)(
f
1
2
g
[
2
] dv
b) and/or by eective Lagrange and/or almost Khler Ricci ows
_
E
_
=
2
_
V
(
s
R+[
f[
2
m
) dv,
S =
_
V
[ (
s
R+[
f[
2
) +
f 2m] dv,
= 2
4
_
V
[[
+
T
f
1
2
g
[
2
] dv,
where all values are constructed equivalently in Cartan and/or almost sym-
plectic variables on /
E
E.
Proof. Similar proofs in coordinate and/or Nadapted forms are given in
[7, 12, 13, 14, 15, 16, 17]. We have to use the corresponding partition function
Z = exp
_
_
V
[
f +m] dv
_
for computations on /
E
E. The formulas in
the conditions of Theorem depend on the type of dconnection,
T,
26
or
T, which is chosen for nonholonomic deformations. Corresponding
re-scaling
f
f, or
f, and , or , have to be considered.
Finally, we note that Ricci ows with dierent dconnections are charac-
terized by dierent thermodynamical values and stationary congurations.
4 Ricci Solitons with Lie Algebroid Symmetries
The problem of constructing distinguished metric and connection struc-
tures encoding generalized symmetries and describing evolution processes
and stationary/xed points is important in geometry and physics. The so
called Ricci solitons and, in particular, (modied) Einstein spaces, or ef-
fective LagrangeFinsler geometries are candidates for constructing exact
solutions. We refer the readers to [37, 38, 7] and references therein for de-
tails on HamiltonPerelman evolution and Ricci solitons with Riemannian
metrics. Recently, some models of Ricci solitons in the semiRiemannian set-
ting have been elaborated with special attention to Lorentzian Lie groups,
algebraic classications and homogeneous metrics, surface projections etc,
see [39, 40, 42, 41, 43, 44].
In this section, we shall construct in explicit form some examples of
exact solutions for Ricci soliton Lie dalgebroid congurations. The st
class of models describes generalized Einstein spaces with nonholonomic (for
instance, almost symplectic) variables and the second one is determined by
LagrangeFinsler generating functions.
4.1 Preliminaries on Lie dalgebroid solitons
Lie dalgebroid Ricci solitons can be viewed as xed points of generalized
Ricci ows with a functional
J (39) satisfying the conditions of Theorem
3.2. Such nonholonomically constrained dynamical systems correspond to
selfsimilar solutions describing Nadapted geometric evolution models.
Denition 4.1 The geometric data [g g,/,
A,
T] [
(, ):= g(
,),
T = +
:] for a complete Riemannian metric g on a smooth T
E
E and
corresponding /
E
E dene a gradient almost KhlerRicci dalgebroid soliton
if there exists a smooth potential function on (x
i
, y
C
) such that
+
T
= g
. (48)
Using the almost symplectic form (27), these equations can be written equiv-
alently in the form
R
[]
+
T
[
T
]
=
.
There are three types of such Ricci solitons determined by = const : steady
ones, for = 0; shrinking, for > 0; and expanding, for < 0.
27
The above classication is important because shrinking solutions for the
Riemannian LeviCivita solitons helps us to understand the asymptotic be-
haviour of ancient solutions of Ricci ows (see, for instance, Proposition 11.2
in [4] and/or Theorem 6.2.1 in [7]). In general, complete gradient shrinking
Ricci solitions describe possible Type I singularity models in the Ricci ow
theory. If = const, the equations (48) transform into the distorted Einstein
equations (36) but for Ricci solitonic congurations.
Proposition 4.1 Let (g g,/,
A,
T; ) be a complete shrinking soliton
on T
E
E and/ or /
E
E. Using nonholonomic frame deformations, we can
construct a redened potential function (x
i
, y
C
), for g g, when (48) are
equivalent to
+
T
= g
. (49)
Proof. Using Conclusion 2.1 and contracting indices in (48), we obtain
that
s
R+ [
T [
2
= const. Distortion relations of type
T
=
T
+
:
allows us to compute
s
R+
s
Z + [(
T +
:) [
2
= const, which can
be rewritten as
s
R + [
T[
2
= const for certain nonlinear transform
. In general, the systems (49) and (48) have dierent solutions. Nev-
ertheless, conditions of type
: = 0 and/or
: = 0 result in the LeviCivita
congurations and equivalent classes of solutions.
4.2 Generalized Einstein equations encoding Lie dalgebroids
We can construct very general classes of odiagonal solutions of (49) if
we impose the condition that in some Nadapted frames
= e
= const, (50)
i.e.
a
= A
a
A
C
a
C
= 0 and 1
A
=
A
.
The information from potential functions is encoded into the data for N
connection structure with coecients A
C
a
.
For simplicity, we shall consider in this section nonholonomic distri-
butions on a nonholonomic E = P with 2 + 2 splitting when a, b, ... =
1, 2; i
, j
= (x
i
, y
a
) = (x
1
, x
2
, y
3
, y
4
). We study nonholonomic
deformations of a dmetric g on E into a target metric g (A.3) on T
E
E,
g g, which results in solutions of the Ricci solitonic equations (49) and
(50). The prime metric is parameterized
g = g
(u)e
=g
i
(x)dx
i
dx
i
+
h
a
(x, y)e
a
e
a
,
for e
= (dx
i
, e
a
= dy
a
+
N
a
i
(u)dx
i
), (51)
e
= (e
i
= /y
a
N
b
i
(u)/y
b
, e
a
= /y
a
).
For physical applications, we can consider that the coecients of such met-
rics are with two Killing vector symmetries and that in certain systems of
28
coordinates it can be diagonalized
8
. In general, we can consider arbitrary
(semi) Riemannian metrics. The target Lie algebroid dmetrics are chosen
g = g
= g
a
A
a
A
a
+ g
A
A
A
(52)
=
a
(x
k
)g
a
A
a
A
a
+
A
(x
k
, y
3
)
h
A
A
,
where we shall construct exact solutions with Killing symmetry on /y
4
(nonKilling congurations request a more advanced geometric techniques).
Let us denote by h
:=
3
and A
3
a
= w
a
(x
k
, y
3
), A
4
a
= n
a
(x
k
, y
3
).
Proposition 4.2 The nontrivial components of the Ricci soliton dalgebroid
equations (49) and (50), with respect to Nadapted bases (17), (18) and for
coordinate transforms when
a
A
a
and 1
A
=
A
for a metric (52, are
R
1
1
=
R
2
2
=
1
2g
1
g
2
[A
1
(A
1
g
2
)
A
1
g
1
A
1
g
2
2g
1
(A
1
g
2
)
2
2g
2
(53)
+A
2
(A
2
g
1
)
A
2
g
1
A
2
g
2
2g
2
(A
2
g
1
)
2
2g
1
] = ,
R
3
3
=
R
4
4
=
1
2h
3
h
4
[h
4
(h
4
)
2
2h
4
3
h
4
2h
3
] = , (54)
R
3a
=
w
a
2h
4
[h
4
(h
4
)
2
2h
4
3
h
4
2h
3
] +
h
4
4h
4
(
A
a
h
3
h
3
+
A
a
h
4
h
4
)
A
a
h
4
2h
4
= 0, (55)
R
4a
=
h
4
2h
3
n
a
+ (
h
4
h
3
h
3
3
2
h
4
)
n
a
2h
3
= 0; (56)
for the equations for the potential function
A
a
w
a
3
n
a
4
= 0 and 1
A
=
A
,
when the torsionless (LeviCivita, LC) conditions
: = 0 transform into
w
a
= (A
a
w
a
3
) ln
_
[h
3
[, (A
a
w
a
3
) ln
_
[h
4
[ = 0, (57)
A
b
w
a
= A
a
w
b
, n
a
= 0,
a
n
b
=
b
n
a
.
Proof. It follows from straightforward computations of the Ricci d
tensor on T
E
E using formulas (A.5), (A.2) and (35). The local frames are
redened in the form
a
= e
a
a
A
a
in order to include Lie algebroid anchor
structure functions and commutation relations of type (14) and (19). For
nonholonomic 2+2+2+...+ decompositions, this can be performed by local
frame/coordinate transforms. Details for calculus on TV are provided in
[18, 17] and references therein. We can work similarly both on TV and T
E
E
but with dierent Nadapted nonholonomic frames and Nelongated partial
derivatives and dierentials. The system of equations (53) - (57) possess
an important decoupling property. For instance, the equation (53) is a 2d
8
this include the bulk of physically important exact solutions of Einstein equations
29
version of Laplace/ dAlambert equation (it depends on signature of the h
metric) with prescribed local source . Such equations can be integrated in
general form even the algebroid structures functions
i
a
(x
k
) are not trivial.
The equation (54) is the same both on TV and T
E
E and contains partial
derivatives only on
3
and can be also in similar form.
4.3 Generating odiagonal solutions
We can integrate the algebroid Ricci soliton equations (49) and (50) for
a nontrivial source , g
a
=
a
e
(x
k
)
,
a
= 1 and h
a
,= 0.
Theorem 4.1 The system (53)(56) decouple in Nadapted form,
1
A
1
(A
1
) +
2
A
2
(A
2
) = 2 (58)
4
= 2h
3
h
4
(59)
w
A
A
= 0, (60)
n
A
+n
A
= 0, (61)
for
A
= h
A
, = h
, =
_
ln [h
4
[
3/2
/[h
3
[
_
, (62)
where = ln [h
4
/
_
[h
3
h
4
[[ (63)
is considered as a generating function.
Proof. It follows from explicit computations for dmetrics (52) with
Killing symmetry on
4
. It is convenient to use the value := e
.
Corollary 4.1 The above systems of nonlinear partial dierential equations,
PDE, can be integrated in very general forms.
Proof. We should follow such a procedure:
1. The (58) is just a 2d Laplace/ dAlamber equation which can be solved
for any given .
2. For h
A
:=
A
z
2
A
(x
k
, y
3
),
A
= 1 (we do not consider summation on
repeating indices in this formula), the system of two equations (59) and
(63) can be written
4
=
3
z
4
(z
3
)
2
and e
z
3
= 2
4
z
4
. Multiplying
both equations for nonzero z
4
,
, z
A
and introducing the result instead
of the rst equation, this system transforms into
= 2
3
4
z
3
z
4
and
z
3
= 2
4
z
4
. Introducing z
3
from the second equation into the rst
one, we obtain [(z
4
)
2
]
=
3
[
2
]
4
4
2
, (64)
30
for an integration function
0
h
4
(x
k
). From the rst equation in above
system, we compute
h
3
=
3
(z
3
)
2
=
4
z
4
z
4
z
4
=
1
2
(ln [[)
(ln [h
4
[)
. (65)
Redening the coordinates and and introducing
3
4
in we express
the solutions in functional form, h
3
[] = (
)
2
/
2
, h
4
[] =
2
/4.
3. To nd w
a
we have to solve certain algebraic equations which can be
obtained if we introduce the coecients (62) in (60),
w
a
= A
a
/
= A
a
/
. (66)
4. Integrating two times on y
3
in (61), we nd
n
b
=
1
n
b
+
2
n
b
_
dy
3
h
3
/(
_
[h
4
[)
3
, (67)
where
1
n
b
(x
i
),
2
n
b
(x
i
) are integration functions.
5. The nonholonomic constraints for the LCconditions (57) can be solved
in explicit form for certain classes of integration functions
1
n
b
and
2
n
b
. We can nd explicit solutions if
2
n
b
= 0 and
1
n
b
= A
b
n with
a function n = n(x
k
). We get (A
a
w
a
3
) 0 for any (x
k
, y
3
)
if w
a
is dened by (66). For any functional H(), we obtain (A
a
w
a
3
)H =
H
(A
a
w
a
3
) = 0. It is possible to solve the equations
(A
a
w
a
3
)h
4
= 0 for h
4
= H([
3
) ln
_
[h
4
[ (A
a
w
a
3
)h
4
.
We can consider a subclass of generating functions =
for which
(A
a
= A
a
(
3
) ln
_
[h
4
[ = 0. The rst system of equa-
tions in (57) can be solved in explicit for any w
a
determined by formulas
(66), and h
3
[
] and h
4
[
]. Let us consider
=
(ln
_
[h
3
[) for a
functional dependence h
3
[
[/[
= A
a
[ ln
_
[h
3
[[/[ ln
_
[h
3
[[
. Taking derivative
3
on
both sides of this equation, we get w
a
=
(Xa| ln
|h
3
||)
| ln
|h
3
||
w
a
| ln
|h
3
||
| ln
|h
3
||
.
The condition w
a
= (A
a
w
a
3
) ln
_
[h
3
[ is necessary for the zero
torsion conditions. It is satised for =
. We can chose w
a
=
w
a
= A
a
= A
a
A, with a nontrivial function
A(x
k
, y
3
) depending
functionally on generating function
in order to solve the equations
A
a
w
b
= A
b
w
a
from the second line in (57).
31
Conclusion 4.1 The class of odiagonal metrics of type (52) with coe-
cients computed following the method outlined in above Proof are determined
by quadratic elements of type ds
2
=
e
(x
k
)
[
1
(A
1
)
2
+
2
(A
2
)
2
] +
3
(
)
2
2
[1
3
+ (A
a
A[
])A
a
]
2
+
4
2
4[[
[1
4
+ (A
a
n)A
a
]
2
.
(68)
In general, on prolongation Lie dalgebroids, the solutions dening Ricci
solitons can be with nontrivial torsion.
Remark 4.1 For arbitrary and related , or
, we can generate o
diagonal solutions of (53)(56) with nonholonomically induced torsion,
ds
2
= e
(x
k
)
[
1
(A
1
)
2
+
2
(A
2
)
2
] + (69)
3
(z
3
)
2
[1
3
+
A
a
A
a
]
2
+
4
(z
4
)
2
[1
4
+ (
1
n
a
+
2
n
a
_
dy
3
(z
3
)
2
(z
4
)
3
)A
a
]
2
,
where the values z
3
(x
k
, y
3
) and z
4
(x
k
, y
3
) are dened by formulas (65) and
(64). In Nadapted frames, the ansatz for such solutions dene a nontrivial
distorting tensor as in
Z =
manifold of di-
mension dimM = n = 2.
Denition 4.2 A Finsler fundamental, or generating, function (metric) is a
function F : TM [0, ) for which 1) F(x, y) is C
on
TM := TM0,
where 0 is the set of zero sections of TM on M; 2) F(x, y) = F(x, y),
for any > 0, i.e. it is a positive 1homogeneous function on the bers of
TM; 3) for any y
T
x
M, the Hessian
v
g
ij
(x, y) =
1
2
2
F
2
y
i
y
j
is considered as
s a vertical (v) metric on typical ber, i.e. it is nondegenerate and positive
denite, det [
v
g
ij
[ , = 0.
For the conditions of Theorem 1.1, we take / = L = F
2
and construct
the geometric data ( g,
N).
32
Theorem 4.2 Any FinslerCartan geometry (equivalently modelled as an
almost KhlerFinsler space) with nonholonomic splitting 2 + 2 can be en-
coded as a canonical Ricci dalgebroid soliton with metric of type (69) and a
respective almost Khler dalgebroid soliton, see Denition 4.1.
Proof. It follows from explicit computations for dmetrics (52) with
Killing symmetry on
4
. We consider for the class of metrics (52) that up
to frame/ coordinate transforms the prime conguration is determined by
a total bundle metric g
= e
(52) de-
nes generic odiagonal solutions for prolongation Lie dalgebroids with
canonical dconnections if g
(, ):= g(
,),
T =
T +:].
The constructions for this theorem can be extended for nonholonomic
splitting of any nite dimension 2+2...+2.
Finally, we note that both primary and target almost Khler Finsler
geometry / Lie algebroid congurations can be quantized, for instance, using
methods of deformation, or Abrane quantization [20, 21].
Acknowledgments: The work is partially supported by the Program
IDEI, PN-II-ID-PCE-2011-3-0256 and contains the main results presented as
a Plenary Lecture at the 11th Panhellenic Geometry Conference (May 31 -
June 2, 2013; Department of Mathematics of the National and Kapodistrian
University of Athens, Greece). The author is grateful to the Organizing
Committee and P. Stavrinos for kind support and collaboration.
A Formulas in Coecient Forms
In this section, we summarize some important local constructions and
coecient formulas which are necessary for formulating Ricci evolution equa-
tions and deriving exact solutions on Lie algebroid models.
A.1 Torsions and curvatures on T
E
P
The Nadapted components
= (L
a
bf
, L
A
Bf
; B
a
bC
, B
A
BC
) of a dconnec-
tion 2.4 and corresponding covariant operator T
= ( e
= (T
)e
. There
are dened the h and vcovariant derivatives, respectively, hT = T
=
33
(L
a
bf
, L
A
Bf
) and vT = T
C
= (B
a
bC
, B
A
BC
), where
L
a
bf
:= (T
f
b
)A
a
, L
A
Bf
:= (T
f
1
B
)
A
, B
a
bC
:= (T
C
b
)A
a
, B
A
BC
:= (T
C
1
B
)
A
are computed for Nadapted bases (17) and (18).
Using rules of absolute dierentiation (14) for Nadapted bases e
:=
, 1
A
and e
:= A
,
B
and the dconnection 1form
:=
,
we can compute the torsion and curvature 2forms of T on T
E
P. For
instance, let us consider such a calculus for the dtorsion 2form. We
take some sections x, y, z of T
E
P parameterized in the form, for example,
z = z
= z
a
a
+z
A
1
A
. Following a Nadapted dierential form calculus,
we prove some important formulas for the dtorsion and dcurvature.
Theorem A.1 For a dconnection T, we can compute
a) the torsion T
:= Te
= de
; the hvcoecients T
=
T
= T
a
bf
, T
a
bA
, T
A
bf
, T
A
Ba
, T
A
BC
with Nadapted coecients
T
a
bf
= L
a
bf
L
a
fb
+C
a
bf
, T
a
bA
= T
a
Ab
= B
a
bA
, T
A
ba
=
A
ba
,
T
A
Ba
=
A
A
a
u
B
L
A
Ba
, T
A
BC
= B
A
BC
B
A
CB
. (A.1)
b) The curvature 1
:= T
= d
= R
,
where R
= e
= R
= R
, R
A
B
, R
A
, R
C
BA
, R
BA
, R
C
BEA
, for
R
a
ebf
=
f
L
a
eb
b
L
a
eb
+L
d
eb
L
a
df
L
d
ef
L
a
db
+ L
a
ed
C
d
bf
B
a
eA
A
fb
,
R
A
Bbf
=
f
L
A
Bb
b
L
A
Bf
+L
C
Bb
L
A
Cf
L
C
Bf
L
A
Cb
+L
A
Bd
C
d
bf
B
A
BC
C
fb
,
R
a
ebA
= 1
A
L
a
eb
T
b
B
a
eA
+B
a
eB
T
B
bA
,
R
C
BfA
= 1
A
L
C
Bf
T
B
C
BA
+B
C
BD
T
D
A
, (A.2)
R
a
bBA
= 1
A
B
a
bB
1
B
B
a
bC
+B
d
bB
B
a
dC
B
d
bC
B
a
dB
,
R
A
ECB
= 1
E
B
A
BC
1
C
B
A
BE
+B
F
BC
B
A
FE
B
F
BE
B
A
FC
.
The formulas (A.1) and (A.2) can be used for T =
T, or T =
T (in the
last case, see respective formulas (A.16) and (A.18)).
A.2 Nadapted coecients for the canonical dconnection
A dmetric structure g = g
on T
E
P is dened by
g = g
= g
ab
A
a
A
b
+ g
AB
A
B
. (A.3)
34
It can be represented in generic odiagonal form,
9
g = g
dz
dz
where
g
=
_
g
ab
+A
A
a
A
B
b
g
AB
A
A
b
g
AC
A
E
a
g
ED
g
DC
_
. (A.4)
Considering dual frame and/or frame transforms, e
dz
and/or
, with e
=
_
e
a
a
A
B
a
e
A
B
0 e
A
A
_
, we obtain quadratic relations be-
tween coecients g
= e
, for
=
(
L
A
B
;
B
C
,
B
A
BC
), where
L
a
bf
=
1
2
g
ae
(
f
g
be
+
b
g
fe
e
g
bf
) +
1
2
g
ae
_
g
bd
C
d
fe
+g
fd
C
d
eb
g
ed
C
d
bf
_
,
L
A
Bf
= 1
B
(A
A
f
) +
1
2
g
AC
_
f
g
BC
g
DC
1
B
(A
D
f
) g
DB
1
C
(A
D
f
)
_
, (A.5)
C
=
1
2
g
1
C
g
,
B
A
BC
=
1
2
g
AD
(1
C
g
BD
+1
B
g
CD
1
D
g
BC
) .
Using such values as Nadapted coecients for a canonical dconnection
T,
we can check that
Tg = 0 and that the h- and v-torsions (A.1) are computed
T
a
bf
= C
a
bf
and
T
A
BC
= 0. There are nontrivial Nadapted coecients
of torsion of
T, i.e.
T
a
bf
,
T
a
bA
,
T
A
ba
and
T
A
Ba
, which can be computed by
introducing the coecients (A.5) into formulas (A.1).
Let us show how we can compute the Nadapted coecients of the
distorting relation
T = +
: in Remark 2.1. Having a dmetric struc-
ture g on T
E
P we can always construct a metric compatible LeviCivita
connection which is completely dened by the zero torsion condition,
= K
= (L
, L
A
B
; B
C
, B
A
BC
), we can verify via straightforward computa-
tions with respect to (17) and (18) that the conditions of the mentioned
Remark us satised by distortion relation
K
+
Z
, (A.6)
where the Nadapted coecients of the distortion d-tensor
: =
are
Z
a
bf
= 0,
Z
A
bf
=
B
a
bB
g
af
g
AB
1
2
A
bf
,
Z
a
Bf
=
1
2
C
af
g
CB
g
ba
ad
bf
B
b
dB
,
Z
A
Bf
=
+
AB
CD
T
C
B
,
Z
A
BC
= 0,
Z
a
fB
=
1
2
A
bf
g
CB
g
ba
+
ad
bf
B
b
dB
,
Z
A
bB
=
AD
CB
T
C
bD
,
Z
a
AB
=
g
ab
2
_
T
C
bA
g
CB
+
T
C
bB
g
CA
_
,
9
i.e. such metrics can not be diagonalized by coordinate transforms
35
for
ad
bf
=
1
2
(
a
b
d
f
g
) and
AB
CD
=
1
2
(
A
C
B
D
g
CD
g
AB
).
Introducing K
, Ricci tensor,
R
,
and scalar curvature,
s
R. The distortions K =
+
Z (A.6) allows us to
compute the distorting tensors (
,
Z
and
s
, R
and
s
R.
A.3 Lie algebroid mechanics and KernMatsumoto models
Let us briey outline some basic constructions [17, 27] when the canoni-
cal N and dconnections and dmetric on T
E
E, for P = E, can be gener-
ated from a regular Lagrangian / as a solution of the corresponding Euler
Lagrange equations. The approach was developed in geometric mechanics
with regular Lagrangians on prolongations of Lie algebroids on bundle maps,
see [33, 30, 32] and references therein (the rst models on mechanics on al-
gebroids were elaborated in [34, 35]).
For a generating function /(x
i
, y
a
) C
E T
E
E and the
Liouville dilaton vector eld (a) =
V
(a, a) = (a, 0, b
V
a
). This allows us to
construct a model of Lie algebroid mechanics for / which can be geometrized
on T
E
E in terms of three geometric objects,
the Cartan 1-section:
L
:= S
(d/) Sec((T
E
E)
);
the Cartan 2-section:
L
:= d
L
Sec(
2
(T
E
E)
); (A.7)
the Lagrangian energy : E
L
:= /i
/ / C
(E),
where the Lie derivative /i
L
= S
(i
X
L
) and i
L
= S
(dE
L
), X Sec(T
E
E). (A.8)
The geometric objects (A.7) and equations (A.8) can are known for var-
ious applications in coecient forms. Using local coordinates (x
i
, y
a
) E
and choosing a basis A
a
, 1
A
Sec(T
E
E), for all a, we have
SA
a
= 1
a
, S1
a
= 0, = y
a
1
a
, E
L
= y
a
//y
a
/, (A.9)
L
=
2
/
y
a
y
b
A
a
1
b
+
1
2
(
i
b
2
/
x
i
y
a
i
a
2
/
x
i
y
b
+C
f
ab
/
y
f
)A
a
A
b
,
36
for Lie algebroid structure functions (
i
a
, C
f
ab
). As a vertical endomorphism
(equivalently, tangent structure) can be used the operator S := A
a
1
a
. A
regular system is characterized by a nondegenerate Hessian
g
ab
:=
2
/
y
a
y
b
, [ g
ab
[ = det [ g
ab
[ , = 0. (A.10)
An EulerLagrange section associated with / is given by any
L
=
y
a
A
a
+
a
1
a
Sec(T
E
E), when functions
a
(x
i
, y
b
) solve this system of
linear equations
b
2
L
y
b
y
a
+ y
b
(
i
b
2
L
x
i
y
a
+ C
f
ab
L
y
f
)
i
a
L
x
i
= 0. The semi
spray vector
e
= g
eb
(
i
b
/
x
i
i
a
2
/
x
i
y
b
y
a
C
f
ba
/
y
f
y
a
) (A.11)
can be found in explicit forms for regular congurations when g
ab
is inverse to
g
ab
. The section
L
transforms into a spray which states that the functions
b
are homogenous of degree 2 on y
b
if the condition [,
L
]
E
=
L
is satised.
The solutions of the EulerLagrange equations for /,
dx
i
d
=
i
a
y
a
and
d
d
(
/
y
a
) +y
b
C
f
ab
/
y
f
i
a
/
x
i
= 0, (A.12)
are parameterized by curves c() = (x
i
(), y
a
()) E.
A.4 The torsion and curvature of the normal dconnection
Let us consider a 1form associated to the normal dconnection
T =
(
T
a
,
T
A
), see
n
T (31) and
= (
L
a
bf
,
B
a
bc
) (32),
a
b
:=
L
a
bf
A
f
+
B
a
bc
c
,
where e
= (e
a
=
a
, 1
A
) and e
:= A
a
,
b
= 1
b
+
A
b
f
1
f
are taken as in
(22). We can prove that the Cartan structure equations are satised,
dA
a
A
b
a
b
= h
T
a
, d
c
c
b
= v
T
c
, (A.13)
and d
a
b
c
b
a
c
=
1
a
b
. (A.14)
The h and vcomponents of the torsion 2form
T
=
_
h
T
a
, v
T
a
_
=
T
a
cb
c
b
from (A.13). The Nadapted coecients are computed
h
T
a
=
B
a
bc
A
b
c
, v
T
a
=
1
2
a
bc
A
b
A
c
+ (1
c
A
a
b
L
a
bc
)A
b
c
, (A.15)
where
a
bc
are coecients of the canonical Nconnection curvature (20) of
the canonical Nconnection, A
b
f
A
b
f
(21). In explicit form, the Nadapted
coecients of dtorsion (A.1) of
T are
T
a
cb
= 0,
T
a
cB
=
B
a
cB
,
T
A
cb
=
A
bc
,
T
A
cB
= 1
B
A
A
c
L
A
cB
,
T
A
CB
= 0, (A.16)
37
where indices A, B, C, ... are transformed into, respectively, a, b, c, ... at the
end (in order to keep the convention on h- and vindices). We note that
we have chosen such nonholonomic distributions when the Y elds from
the Theorem 2.5 are such way stated that the formulas (A.16) on T
E
E are
symilar to those on TM.
Using a dierential form calculus, we can also compute the curvature
2form (A.14),
1
e
f
=
R
e
fab
b
=
1
2
1
e
fab
A
a
A
b
+
T
e
faB
A
a
B
+
1
2
o
e
fAB
B
, (A.17)
where the nontrivial Nadapted coecients are
1
e
fab
=
b
L
e
fa
L
e
fb
+
L
d
fa
L
e
db
L
d
fb
L
e
da
B
e
fA
A
ba
, (A.18)
T
a
bfB
= 1
B
L
a
bf
T
f
B
a
bB
,
o
e
fAB
= 1
B
B
e
fA
1
A
B
e
fB
+
B
d
fA
B
e
dB
B
d
fB
B
e
dA
.
The distortion relations for the Ricci tensor are computed
R
=
R
+
Z
, (A.19)
R
= R
= e
+K
+K
=
Z
= e
+
Z
+
Z
+
Z
.
References
[1] R. S. Hamilton, J. Di. Geom. 17 (1982) 255
[2] R. S. Hamilton, in Surveys in Dierential Geometry, Vol. 2 (Interna-
tional Press, 1995), pp. 7
[3] R. S. Hamilton, Contemp. Math. 71 (1988) 237
[4] G. Perelman, arXiv: math.DG/ 0211159
[5] G. Perelman, arXiv: math. DG/ 0303109
[6] G. Perelman, arXiv: math.DG/ 0307245
[7] H. -D. Cao and X. -P. Zhu, Asian J. Math., 10 (2006) 165
[8] B. Kleiner and J. Lott, Geometry & Topology 12 (2008) 2587
[9] J. W. Morgan and G. Tian, Ricci ow and the Poincar conjecture,
AMS, Clay Mathematics Monographs, vol. 3 (2007)
[10] A. Chau, J. Di. Geom. 66 (2004) 211-232
38
[11] J. Streets and G. Tian, Symplectic curvature ow, arXiv: 1012.2104
[12] S. Vacaru, J. Math. Phys. 49 (2008) 043504
[13] S. Vacaru, Rep. Math. Phys. 63 (2009) 95
[14] S. Vacaru, Int. J. Geom. Methods. Mod. Phys. 9 (2012) 120041
[15] S. Vacaru, J. Math. Phys. 50 (2009) 073503
[16] S. Vacaru, Chaos, Solitons & Fractals 45 (2012) 1266
[17] S. Vacaru, Lagrange-Ricci Flows and Evolution of Geometric Mechanics
and Analogous Gravity on Lie Algebroids, arXiv: 1108.4333
[18] S. Vacaru, Int. J. Geom. Methods. Mod. Phys. 5 (2008) 473
[19] A. Karabegov and M. Schlichenmaier, Lett. Math. Phys. 57 (2001) 135
[20] S. Vacaru, J. Geom. Phys. 60 (2010) 1289
[21] S. Vacaru, Int. J. Geom. Meth. Mod. Phys. 6 (2009) 873
[22] C. Ehresmann, Coloque de Topologie, Bruxelles (1955) 29
[23] J. Kern, Archiv der Mathematik (Basel) 25 (1974) 438
[24] K. Yano and S. Ishihara, Tangent and Cotangent Bundles: Dierential
Geometry (Marcel Dekker Inc., NY, 1973)
[25] M. Matsumoto, Foundations of Finsler Geometry and Special Finsler
Spaces (Kaisisha: Shingaken, Japan, 1986)
[26] S. Vacaru, J. Math. Phys. 47 (2006) 093504
[27] S. Vacaru, Mathematical Sciences [Springer Open] 6 (2012) 18; arXiv:
0705.0032
[28] P. J. Higgins, K. Mackenzie, J. Algebra, 129 (1990) 194
[29] K. Mackenzie, General Theory of Lie Groupoids and Lie Algebroids,
Londom Mathematical Society Lecture Note Series: 213 (Cambridge
University Press, 2005)
[30] E. Martnez, Acta Appl. Math. 67 (2001) 295
[31] E. Martnez, T. Mesdag and W. Sarlet, J. Geom. Phys. 44 (2002) 70
[32] M. de Len, J. C. Marrero and E. Martnez, J. Phys. A: Math. Gen. 38
(2005) R241
39
[33] J. Corts, M. de Len, J. C. Marrero, D. Martin de Diego and E.
Martnez, arXiv: math-ph/0511009
[34] A. Weinstein, Fields Inst. Commun. 7 (1996) 207
[35] P. Libermann, Arch. Math. (Brno) 32 (1996) 147
[36] S. Vacaru, Int. J. Theor. Phys. 48 (2009) 579
[37] H. -D. Cao, R. S. Hamilton and T. Ilmanen, Gaussian densities and
stability for some Ricci solitons, arXiv: math.DG/ 0404165
[38] H. -D. Cao, D. Zhou, J. Di. Geom 85 (2010) 175-185
[39] J. Lauret, Math. Ann. 319 (2001) 715-733
[40] P. Baird and L. Danieleo, J. Reine Angew. Math. 608 (2007) 65-91
[41] M. BrozosVazquez, E. GarciaRio and S. GavinoFernandez, Locally
conformally at Lorentzian gradient Ricci solitons, arXiv: 1106.2924
[42] W. Batat and K. Onda, Algebraic Ricci solitons of threediemnsional
Lorentian Lie groups, arXiv: 1112.2455
[43] G. Calvaruso and A. Fino, Fourdimensional pseudo-Riemannian ho-
mogeneous Ricci solitons, arXiv: 1111.6384
[44] M. Jablonski, Homogeneous Ricci solitons, arXiv: 1109.6556
40