A N N A L E S
U N I V E R S I T A T I S M A R I A E C U R I E - S K Ł O D O W S K A L U B L I N – P O L O N I A
VOL. LXIX, NO. 1, 2015 SECTIO A 91–108
MARIUSZ PLASZCZYK
The natural transformations between r-th order prolongation of tangent and cotangent bundles
over Riemannian manifolds
Abstract. If (M, g) is a Riemannian manifold then there is the well-known base preserving vector bundle isomorphism T M→ T∗M given by v → g(v, −) between the tangent T M and the cotangent T∗M bundles of M. In the present note first we generalize this isomorphism to the one JrT M → JrT∗M between the r-th order prolongation JrT M of tangent T M and the r-th order prolongation JrT∗M of cotangent T∗M bundles of M. Further we describe all base preserving vector bundle maps DM(g): JrT M → JrT∗M depending on a Riemannian metric g in terms of natural (in g) tensor fields on M .
1. Introduction. All manifolds are smooth, Hausdorff, finite dimensional and without boundaries. Maps are assumed to be smooth, i.e. of class C
∞. Let Mf
mdenote category of m-dimensional manifolds and their embed- dings.
From the general theory it is well known that the tangent T M and the cotangent T
∗M bundles of M are not canonically isomorphic. However, if g is a Riemannian metric on a manifold M , there is the base preserving vector bundle isomorphism i
g: T M → T
∗M given by i
g(v) = g(v, −), v ∈ T
xM, x ∈ M.
In the second section of the present note we give necessary definitions.
2010 Mathematics Subject Classification. 58A05, 58A20, 58A32.
Key words and phrases. Riemannian manifold, higher order prolongation of a vector bundle, natural tensor, natural operator.
In the third section first we generalize the isomorphism i
g: T M → T
∗M depending on g to a base preserving vector bundle isomorphism J
ri
g: J
rT M
→ J
rT
∗M canonically depending on g between the r-th order prolongation J
rT M of tangent T M and the r-th order prolongation J
rT
∗M of cotan- gent T
∗M bundles of M . Next we construct another more advanced base preserving vector bundle isomorphism i
<r>g: J
rT M → J
rT
∗M canonically depending on g.
In the fourth section we consider the problem of describing all Mf
m- natural operators D : Riem Hom(J
rT, J
rT
∗) transforming Riemannian metrics g on m-dimensional manifolds M into base preserving vector bundle maps D
M(g): J
rT M → J
rT
∗M . Our studies lead to the reduction of this problem to the one of describing all Mf
m-natural operators t : Riem T
∗⊗ S
lT ⊗ T
∗⊗ S
kT
∗(for l, k = 1, . . . , r) sending Riemannian metrics g on M into tensor fields t
M(g) of types T
∗⊗ S
lT ⊗ T
∗⊗ S
kT
∗.
2. Definitions. Now we give some necessary definitions.
Definition 1. The r-th order prolongation of tangent bundle is a functor J
rT : Mf
m→ VB sending any m-manifold M into J
rT M and any embed- ding ϕ : M
1→ M
2of two manifolds into J
rT ϕ : J
rT M
1→ J
rT M
2given by J
rT ϕ (j
xrX ) = j
ϕ(x)rϕ
∗X, where X ∈ X (M
1) and ϕ
∗X = T ϕ ◦ X ◦ ϕ
−1is the image of a vector field X by ϕ.
Definition 2. The r-th order prolongation of cotangent bundle is a func- tor J
rT
∗: Mf
m→ VB sending any m-manifold M into J
rT
∗M and any embedding ϕ : M
1→ M
2of two manifolds into
J
rT
∗ϕ : J
rT
∗M
1→ J
rT
∗M
2given by J
rT
∗ϕ := J
r(T ϕ
−1)
∗.
Definition 3. The dual bundle of the r-th order prolongation of tangent bundle is a functor (J
rT )
∗: Mf
m→ VB sending any m-manifold M into (J
rT )
∗M := (J
rT M )
∗and any embedding ϕ : M
1→ M
2of two manifolds into
(J
rT )
∗ϕ : (J
rT )
∗M
1→ (J
rT )
∗M
2given by (J
rT )
∗ϕ := (J
rT ϕ
−1)
∗.
Definition 4. The dual bundle of the r-th order prolongation of cotangent bundle is a functor (J
rT
∗)
∗: Mf
m→ VB sending any m-manifold M into (J
rT
∗)
∗M := (J
rT
∗M )
∗and any embedding ϕ : M
1→ M
2of two manifolds into
(J
rT
∗)
∗ϕ : (J
rT
∗)
∗M
1→ (J
rT
∗)
∗M
2given by (J
rT
∗)
∗ϕ := (J
rT
∗ϕ
−1)
∗.
The general concept of natural operators can be found in [4]. In partic-
ular, we have the following definitions.
Definition 5. An Mf
m-natural operator D : Riem Hom(J
rT, J
rT
∗) transforming Riemannian metrics g on m-dimensional manifolds M into base preserving vector bundle maps D
M(g): J
rT M → J
rT
∗M is a system D = {D
M}
M∈obj(Mfm)of regular operators
D
M: Riem(M) → Hom
M(J
rT M, J
rT
∗M )
satisfying the Mf
m-invariance condition, where Hom
M(J
rT M, J
rT
∗M ) is the set of all vector bundle maps J
rT M → J
rT
∗M covering the identity map id
Mof M .
The Mf
m-invariance condition of D is following: for any g
1∈ Riem(M
1) and g
2∈ Riem(M
2) if g
1and g
2are ϕ-related by an embedding ϕ : M
1→ M
2of m-manifolds (i.e. ϕ is (g
1, g
2)-isomorphism) then D
M1(g
1) and D
M2(g
2) are also ϕ-related (i.e. D
M2(g
2) ◦ J
rT ϕ = J
rT
∗ϕ ◦ D
M1(g
1)).
Equivalently, the above Mf
m-invariance means that for any g
1∈ Riem (M
1) and g
2∈ Riem(M
2) if the diagram
(1)
T
∗M
1⊗ T
∗M
1T
∗M
2⊗ T
∗M
2M
1M
2T
∗ϕ ⊗ T
∗ϕ
g
1ϕ
g
2commutes for an embedding ϕ : M
1→ M
2(i.e. (T
∗ϕ ⊗ T
∗ϕ ) ◦ g
1= g
2◦ ϕ) then the diagram
J
rT
∗M
1J
rT
∗M
2J
rT M
1J
rT M
2J
rT
∗ϕ
D
M1(g
1)
J
rT ϕ
D
M2(g
2)
commutes also.
We say that operator D
Mis regular if it transforms smoothly parame- terized families of Riemannian metrics into smoothly parameterized ones of vector bundle maps.
Similarly, we can define the following concepts:
- an Mf
m-natural operator D : Riem Hom(J
rT, J
rT ),
- an Mf
m-natural operator D : Riem Hom(J
rT, (J
rT )
∗),
- an Mf
m-natural operator D : Riem Hom(J
rT, (J
rT
∗)
∗),
- an Mf
m-natural operator D : Riem Hom(J
rT
∗, J
rT ), - an Mf
m-natural operator D : Riem Hom(J
rT
∗, J
rT
∗), - an Mf
m-natural operator D : Riem Hom(J
rT
∗, (J
rT )
∗), - an Mf
m-natural operator D : Riem Hom(J
rT
∗, (J
rT
∗)
∗), - an Mf
m-natural operator D : Riem Hom((J
rT )
∗, J
rT ), - an Mf
m-natural operator D : Riem Hom((J
rT )
∗, J
rT
∗), - an Mf
m-natural operator D : Riem Hom((J
rT )
∗, (J
rT )
∗), - an Mf
m-natural operator D : Riem Hom((J
rT )
∗, (J
rT
∗)
∗), - an Mf
m-natural operator D : Riem Hom((J
rT
∗)
∗, J
rT ), - an Mf
m-natural operator D : Riem Hom((J
rT
∗)
∗, J
rT
∗), - an Mf
m-natural operator D : Riem Hom((J
rT
∗)
∗, (J
rT )
∗), - an Mf
m-natural operator D : Riem Hom((J
rT
∗)
∗, (J
rT
∗)
∗).
Now we have the following definition.
Definition 6. An Mf
m-natural operator A : Riem (T ⊗S
lT
∗, T
∗⊗S
kT
∗) transforming Riemannian metrics g on m-dimensional manifolds M into base preserving vector bundle maps A
M(g): T M⊗S
lT
∗M → T
∗M ⊗S
kT
∗M is a system A = {A
M}
M∈obj(Mfm)of regular operators A
M: Riem(M) → C
∞(T M ⊗ S
lT
∗M, T
∗M ⊗ S
kT
∗M ) satisfying the Mf
m-invariance condi- tion, where C
∞(T M ⊗S
lT
∗M, T
∗M ⊗S
kT
∗M ) is the set of all vector bundle maps T M ⊗S
lT
∗M → T
∗M ⊗S
kT
∗M covering the identity map id
Mof M . The Mf
m-invariance condition of A is following : for any g
1∈ Riem(M
1) and g
2∈ Riem(M
2) if g
1and g
2are ϕ-related by an embedding ϕ : M
1→ M
2of m-manifolds (i.e. (T
∗ϕ ⊗T
∗ϕ )◦g
1= g
2◦ϕ) then A
M1(g
1) and A
M2(g
2) are also ϕ-related (i.e. A
M2(g
2)◦(T ϕ⊗S
lT
∗ϕ ) = (T
∗ϕ ⊗S
kT
∗ϕ )◦A
M1(g
1)).
Equivalently, the above Mf
m-invariance means that for any g
1∈ Riem (M
1) and g
2∈ Riem(M
2) if the diagram (1) commutes for an em- bedding ϕ : M
1→ M
2then the diagram
T
∗M
1⊗ S
kT
∗M
1T
∗M
2⊗ S
kT
∗M
2T M
1⊗ S
lT
∗M
1T M
2⊗ S
lT
∗M
2T
∗ϕ ⊗ S
kT
∗ϕ
A
M1(g
1)
T ϕ ⊗ S
lT
∗ϕ
A
M2(g
2)
commutes also.
The regularity means almost the same as in Definition 5.
Similarly, we can define the following concepts:
- an Mf
m-natural operator A : Riem (T ⊗ S
lT, T ⊗ S
kT ),
- an Mf
m-natural operator A : Riem (T
∗⊗ S
lT, T ⊗ S
kT ),
- an Mf
m-natural operator A : Riem (T ⊗ S
lT
∗, T ⊗ S
kT ), - an Mf
m-natural operator A : Riem (T ⊗ S
lT, T
∗⊗ S
kT ), - an Mf
m-natural operator A : Riem (T ⊗ S
lT, T ⊗ S
kT
∗), - an Mf
m-natural operator A : Riem (T
∗⊗ S
lT
∗, T ⊗ S
kT ), - an Mf
m-natural operator A : Riem (T
∗⊗ S
lT, T
∗⊗ S
kT ), - an Mf
m-natural operator A : Riem (T
∗⊗ S
lT, T ⊗ S
kT
∗), - an Mf
m-natural operator A : Riem (T ⊗ S
lT
∗, T
∗⊗ S
kT ), - an Mf
m-natural operator A : Riem (T ⊗ S
lT
∗, T ⊗ S
kT
∗), - an Mf
m-natural operator A : Riem (T ⊗ S
lT, T
∗⊗ S
kT
∗), - an Mf
m-natural operator A : Riem (T
∗⊗ S
lT
∗, T
∗⊗ S
kT ), - an Mf
m-natural operator A : Riem (T
∗⊗ S
lT
∗, T ⊗ S
kT
∗), - an Mf
m-natural operator A : Riem (T
∗⊗ S
lT, T
∗⊗ S
kT
∗), - an Mf
m-natural operator A : Riem (T
∗⊗ S
lT
∗, T
∗⊗ S
kT
∗).
Next we have an important general definition of natural tensor.
Definition 7. An Mf
m-natural operator (natural tensor) t : Riem
pT
⊗
qT
∗transforming Riemannian metrics g on m-dimensional manifolds M into tensor fields of type (p, q) on M is a system t = {t
M}
M∈obj(Mfm)of regular operators t
M: Riem(M) → T
(p,q)(M) satisfying the Mf
m-invar- iance condition, where T
(p,q)(M) is the set of tensor fields of type (p, q) on M .
The Mf
m-invariance condition of t is following : for any g
1∈ Riem(M
1) and g
2∈ Riem(M
2) if g
1and g
2are ϕ-related by an embedding ϕ : M
1→ M
2of m-manifolds (i.e. (T
∗ϕ ⊗T
∗ϕ )◦g
1= g
2◦ϕ) then t
M1(g
1) and t
M2(g
2) are also ϕ-related (i.e. t
M2(g
2) ◦ ϕ = (
pT ϕ ⊗
qT
∗ϕ ) ◦ t
M1(g
1)).
Equivalently, the above Mf
m-invariance means that for any g
1∈ Riem (M
1) and g
2∈ Riem(M
2) if the diagram (1) commutes for an em- bedding ϕ : M
1→ M
2, then the diagram
pT M
1⊗
qT
∗M
1 pT M
2⊗
qT
∗M
2M
1M
2 pT ϕ ⊗
qT
∗ϕ
t
M1(g
1)
ϕ
t
M2(g
2)
commutes also.
We say that operator t
Mis regular if it transforms smoothly parametrized families of Riemannian metrics into smoothly parametrized ones of tensor fields.
Now we have a definition of a special kind of natural tensor.
Definition 8. An Mf
m-natural operator (natural tensor) t : Riem T
∗⊗ S
lT ⊗ T
∗⊗ S
kT
∗transforming Riemannian metrics g on m-dimensional manifolds M into tensor fields of type T
∗⊗S
lT ⊗T
∗⊗S
kT
∗on M is a system t = {t
M}
M∈obj(Mfm)of regular operators t
M: Riem(M) → C
∞(T
∗M ⊗ S
lT M ⊗ T
∗M ⊗ S
kT
∗M ) satisfying the Mf
m-invariance condition, where C
∞(T
∗M ⊗ S
lT M ⊗ T
∗M ⊗ S
kT
∗M ) is the set of all tensor fields of type T
∗⊗ S
lT ⊗ T
∗⊗ S
kT
∗on M .
The Mf
m-invariance condition of t is following: for any g
1∈ Riem(M
1) and g
2∈ Riem(M
2) if g
1and g
2are ϕ-related by an embedding ϕ : M
1→ M
2of m-manifolds (i.e. (T
∗ϕ ⊗T
∗ϕ )◦g
1= g
2◦ϕ), then t
M1(g
1) and t
M2(g
2) are also ϕ-related (i.e. t
M2(g
2)◦ϕ = (T
∗ϕ ⊗S
lT ϕ ⊗T
∗ϕ ⊗S
kT
∗ϕ )◦t
M1(g
1)).
Equivalently, the above Mf
m-invariance means that for any g
1∈ Riem (M
1) and g
2∈ Riem(M
2) if the diagram (1) commutes for an em- bedding ϕ : M
1→ M
2, then the diagram
T
∗M
1⊗S
lT M
1⊗T
∗M
1⊗S
kT
∗M
1T
∗M
2⊗S
lT M
2⊗T
∗M
2⊗S
kT
∗M
2M
1M
2Φ
t
M1(g
1)
ϕ
t
M2(g
2)
commutes also, where Φ = T
∗ϕ ⊗ S
lT ϕ ⊗ T
∗ϕ ⊗ S
kT
∗ϕ.
The regularity means almost the same as in Definition 7.
Similarly, we can define the following concepts:
- an Mf
m-natural operator (natural tensor) t : Riem T ⊗ S
lT ⊗ T ⊗ S
kT ,
- an Mf
m-natural operator (natural tensor) t : Riem T
∗⊗ S
lT ⊗ T ⊗ S
kT ,
- an Mf
m-natural operator (natural tensor) t : Riem T ⊗ S
lT
∗⊗ T ⊗ S
kT ,
- an Mf
m-natural operator (natural tensor) t : Riem T ⊗ S
lT ⊗ T
∗⊗ S
kT ,
- an Mf
m-natural operator (natural tensor) t : Riem T ⊗ S
lT ⊗ T ⊗ S
kT
∗,
- an Mf
m-natural operator (natural tensor) t : Riem T
∗⊗ S
lT
∗⊗ T ⊗ S
kT ,
- an Mf
m-natural operator (natural tensor) t : Riem T
∗⊗ S
lT ⊗
T
∗⊗ S
kT ,
- an Mf
m-natural operator (natural tensor) t : Riem T
∗⊗ S
lT ⊗ T ⊗ S
kT
∗,
- an Mf
m-natural operator (natural tensor) t : Riem T ⊗ S
lT
∗⊗ T
∗⊗ S
kT ,
- an Mf
m-natural operator (natural tensor) t : Riem T ⊗ S
lT
∗⊗ T ⊗ S
kT
∗,
- an Mf
m-natural operator (natural tensor) t : Riem T ⊗ S
lT ⊗ T
∗⊗ S
kT
∗,
- an Mf
m-natural operator (natural tensor) t : Riem T
∗⊗ S
lT
∗⊗ T
∗⊗ S
kT ,
- an Mf
m-natural operator (natural tensor) t : Riem T
∗⊗ S
lT
∗⊗ T ⊗ S
kT
∗,
- an Mf
m-natural operator (natural tensor) t : Riem T ⊗ S
lT
∗⊗ T
∗⊗ S
kT
∗,
- an Mf
m-natural operator (natural tensor) t : Riem T
∗⊗ S
lT
∗⊗ T
∗⊗ S
kT
∗.
In the third section we present also explicit examples of Mf
m-natural operators D : Riem Hom(J
rT, J
rT
∗).
A full description of all polynomial natural tensors t : Riem
pT ⊗
qT
∗transforming Riemannian metrics on m-manifolds into tensor fields of types (p, q) can be found in [1]. This description is following. Each covariant derivative of the curvature R(g) ∈ T
(0,4)(M) of a Riemannian metric g is a natural tensor and the metric g is also a natural tensor. Further all the natural tensors t : Riem
pT ⊗
qT
∗can be obtained by a procedure:
(a) every tensor multiplication of two natural tensors give a new natural tensor,
(b) every contraction on one covariant and one contravariant entry of a natural tensor give a new natural tensor,
(c) we can tensorize any natural tensor with a metric independent nat- ural tensor,
(d) we can permute any number of entries in the tensor product, (e) we can repeat these steps,
(f) we can take linear combinations.
Furthermore, if we take respective type natural tensors and apply respec- tive symmetrization, then we can produce many natural tensors t : Riem T
∗⊗ S
lT ⊗ T
∗⊗ S
kT
∗.
3. Constructions.
Example 1. Let (M, g) be a Riemannian manifold. Then we have a base preserving vector bundle isomorphism i
g: T M → T
∗M given by
i
g(v) = g(v, −), v ∈ T
xM, x ∈ M.
Next we can obtain a base preserving vector bundle isomorphism J
ri
g: J
rT M → J
rT
∗M defined by a formula
J
ri
g(j
rxX ) = j
xr(i
g◦ X),
where X ∈ X (M). Similarly we receive also a base preserving vector bundle isomorphism
(J
ri
−1g)
∗: (J
rT M )
∗→ (J
rT
∗M )
∗.
Because of the canonical character of the above constructions we get the following propositions.
Proposition 1. The family A
(r): Riem Hom(J
rT, J
rT
∗) of operators A
(r)M: Riem(M) → Hom
M(J
rT M, J
rT
∗M ), A
(r)M(g) = J
ri
gfor all M ∈ obj(Mf
m) is an Mf
m-natural operator.
Proposition 2. The family A
[r]: Riem Hom((J
rT )
∗, (J
rT
∗)
∗) of oper- ators
A
[r]M: Riem(M) → Hom
M((J
rT M )
∗, (J
rT
∗M )
∗), A
[r]M(g) = (J
ri
−1g)
∗for all M ∈ obj(Mf
m) is an Mf
m-natural operator.
Now we are going to present another more advanced example of an Mf
m- natural operator D : Riem Hom(J
rT, J
rT
∗).
Recall that if g is a Riemannian tensor field on a manifold M and x ∈ M , then there is g-normal coordinate system ϕ : (M, x) → (R
m, 0) with centre x. If ψ : (M, x) → (R
m, 0) is another g-normal coordinate system with centre x, then there is A ∈ O(m) such that ψ = A ◦ ϕ near x. Let I : J
0rT R
m→
rk=1
T
0R
m⊗ S
kT
0∗R
m=
rk=0
R
m⊗ S
kR
m∗(see [3]) and I
1: J
0rT
∗R
m→
rk=1
T
0∗R
m⊗ S
kT
0∗R
m=
rk=0
R
m∗⊗ S
kR
m∗(see [7]) be the standard O (m)-invariant vector space isomorphisms.
We have the following important proposition.
Proposition 3. Let g be a Riemannian tensor field on a manifold M . Then there are (canonical in g) vector bundle isomorphisms
I
g: J
rT M →
r k=0T M ⊗ S
kT
∗M,
J
g: J
rT
∗M →
r k=0T
∗M ⊗ S
kT
∗M,
(I
g−1)
∗: (J
rT M )
∗→
r
k=0
T M ⊗ S
kT
∗M
∗∼ =
r k=0T
∗M ⊗ S
kT M,
(J
g−1)
∗: (J
rT
∗M )
∗→
r
k=0
T
∗M ⊗ S
kT
∗M
∗∼ =
r k=0T M ⊗ S
kT M.
Proof. Let v = j
xrX ∈ J
xrT M , where X ∈ X (M), x ∈ M. Let ϕ: (M, x) → (R
m, 0) be a g-normal coordinate system with centre x. We define
I
g(v) := I
gϕ(v) =
r
k=0
T ϕ
−1⊗ S
kT
∗ϕ
−1◦ I ◦ J
rT ϕ (v).
If ψ : (M, x) → (R
m, 0) is another g-normal coordinate system with centre x, then ψ = A ◦ ϕ (near x) for some A ∈ O(m). The O(m)-invariance of I means that
(2) I ◦ J
rT A =
r
k=0
T
0A ⊗ S
kT
0∗A
◦ I.
Hence we deduce that I
gψ(v) =
r
k=0
T ψ
−1⊗ S
kT
∗ψ
−1◦ I ◦ J
rT ψ (v)
=
r k=0(T (A ◦ ϕ)
−1⊗ S
kT
∗(A ◦ ϕ)
−1) ◦ I ◦ J
rT (A ◦ ϕ)(v)
=
r k=0((T ϕ
−1◦ T A
−1) ⊗ S
kT
∗(ϕ
−1◦ A
−1)) ◦ I ◦ (J
rT A ◦ J
rT ϕ )(v)
=
r k=0((T ϕ
−1◦ T A
−1) ⊗ S
kT
∗(ϕ
−1◦ A
−1)) ◦ (I ◦ J
rT A ) ◦ J
rT ϕ (v) =: L.
Now using (2), we receive L =
r k=0((T ϕ
−1◦T A
−1)⊗S
kT
∗(ϕ
−1◦A
−1))◦
r
k=0
T A ⊗S
kT
∗A
◦I ◦J
rT ϕ (v)
=
r k=0((T ϕ
−1◦T A
−1)◦T A) ⊗ (S
kT
∗(ϕ
−1◦A
−1)◦S
kT
∗A )
◦I ◦J
rT ϕ (v)
=
r k=0((T ϕ
−1◦ T A
−1◦ T A) ⊗ S
kT
∗(ϕ
−1◦ A
−1◦ A)) ◦ I ◦ J
rT ϕ (v)
=
r k=0(T ϕ
−1⊗ S
kT
∗ϕ
−1) ◦ I ◦ J
rT ϕ (v) = I
gϕ(v).
Therefore, the definition of I
g(v) is independent of the choice of ϕ. So, isomorphism I
g: J
rT M →
rk=0
T M ⊗ S
kT
∗M is well defined.
Similarly, we put J
g(v) := J
gϕ(v) =
r
k=0
T
∗ϕ
−1⊗ S
kT
∗ϕ
−1◦ I
1◦ J
rT
∗ϕ (v).
Using O (m)-invariance of I
1(i.e. I
1◦ J
rT
∗A = (
rk=0
T
0∗A ⊗ S
kT
0∗A ) ◦ I
1) analogously as before, we show that I
gψ(v) = I
gϕ(v). This proves that the definition of J
g(v) is independent of the choice of g-normal coordinate system ϕ with centre x and the isomorphism J
g: J
rT
∗M →
rk=0
T
∗M ⊗ S
kT
∗M is well defined.
Finally we obtain (canonical in g) vector bundle isomorphisms (I
g−1)
∗: (J
rT M )
∗→
r
k=0
T M ⊗ S
kT
∗M
∗∼ =
r k=0T
∗M ⊗ S
kT M
(J
g−1)
∗: (J
rT
∗M )
∗→
r
k=0
T
∗M ⊗ S
kT
∗M
∗∼ =
r k=0T M ⊗ S
kT M. Remark 1. W. Mikulski (in [7]) has recently constructed a (canonical in
∇) vector bundle isomorphism I
∇: J
rT M →
rk=0
T
∗M ⊗ S
kT
∗M for a classical linear connection ∇ on a manifold M.
Now we have further important identifications.
Example 2. Let (M, g) be a Riemannian manifold and i
g: T M → T
∗M be a base preserving vector bundle isomorphism recalled in Example 1. Using the base preserving vector bundle isomorphisms I
gand J
gfrom Proposi- tion 3, we receive the following vector bundle isomorphisms
i
<r>g:= J
g−1◦
r
k=0
i
g⊗ S
kT
∗id
M◦ I
g: J
rT M → J
rT
∗M,
i
[r]g:= J
g∗◦
r
k=0