Professional Documents
Culture Documents
Euler-Lagrange-Hamilton Dynamics
with Fractional Action
CONSTANTIN UDRISTE DUMITRU OPRIS
University Politehnica of Bucharest West University of Timisoara
Department of Mathematics Department of Applied Mathematics
Splaiul Indpendentei 313 B-dul Vasile Parvan 4
060042, Bucharest 1900 Timisoara
ROMANIA ROMANIA
udriste@mathem.pub.ro opris@math.uvt.ro
Abstract: Our aim is three-fold: to point out that the fractional integral actions are coming from Stieltjes actions, to
find the roots and the geometry of some Euler-Lagrange or Hamilton ODEs or PDEs, to evidentiate some ideas that
include the fractal theory of solids. Section 1 discusses the Euler-Lagrange ODEs associated to single-time Stieltjes
actions. Teir dual Hamilton ODEs are analized in Section 2. Section 3 studies the geometry associated to single-
time Euler-Lagrange or Hamilton operators. Section 4 analyzes the Euler-Lagrange PDEs associated to multitime
Stieltjes actions (multiple or curvilinear integrals). Section 5 formulates the multitime perimetric problem of non-
renewable resources. Section 6 studies the Hamilton PDEs associated to multitime Stieltjes actions. Section 7
describes the geometry associated to multitime Euler-Lagrange or Hamilton operators (dynamical connection and
semi-spray, Poincaré-Cartan form, Hamilton-Poisson systems on jet bundle). Section 8 formulates a multitime
Hamilton-Poisson systems theory on jet bundle.
Key–Words: fractional Stieltjes action, Euler-Lagrange or Hamilton equations, dynamic connection, symplectic
manifold.
The single-time Stieltjes action is defined via the Stiel- and a differentiable function gτ : R → R. Then
jes integral of L with respect to gτ (t) in the sense of the Euler-Lagrange ODE associated to the Lagrangian
functional L = L(x, ẋ)gτ0 (t) is
Z τ
Iτ (x(·)) = L(t, x(t), ẋ(t))dgτ (t). gτ00 (t)
0 ẍ − x − (ẋ − ax) = 0.
gτ0 (t)
Particularly, we define the single-time action of
L(t, x(t), xα (t)) with respect to the weight gτ0 (t) by If the function gτ is given by (1), then
the Riemann integral
r−1 r−1
Z τ ẍ − ( a + 1)x − ẋ = 0;
τ −t τ −t
Iτ (x(·)) = L(t, x(t), ẋ(t))gτ0 (t)dt, (4)
0
if gτ is given by (2), then ẍ − τ ẋ + (1 + aτ )x = 0.
where τ is fixed. The function L(t, x(t), ẋ(t)) = Remarks. 1) A particular weight gτ0 (t) can be ob-
L(t, x(t), ẋ(t))gτ0 (t) is called Lagrangian. tained taking the Riemannian manifold (R, hτ (t) >
Examples. 1) The fractional action from physics 0) instead the Euclidean manifold (R, 1).pIn this case
τ the Lagrangian is L = L(t, x(t), ẋ(t)) hτ (t) and
1
Z
gτ0 (t) = hτ (t).
p
Iτ (x(·)) = L(t, x(t), ẋ(t))(τ − t)r−1 dt
Γ(r) 0 2) If we have in mind only the Lagrangian density
g 00 (t) ∂L
obtained for the function gτ (t) in (1). Particularly, for L, then the term Fi = τ0 in Euler-Lagrange
r = 1 we obtain the classical action. gτ (t) ∂ ẋi
2) The discounted action at rate τ from eco- ODEs (4) stands for an external force.
nomics 3) If the function gτ (t) is given by (1), then the
ODEs (4) reduce to
Z τ
Iτ (x(·)) = L(t, x(t), ẋ(t))e−τ t dt ∂L d ∂L 1 − r ∂L
0 − = , i = 1, ..., n.
∂xi dt ∂ ẋi τ − t ∂ ẋi
obtained for gτ (t) in (2).
3) The fractal action from physics [11] In particular, for r = 1 we obtain the classical Euler-
Z τ Lagrange ODEs.
Iτ (x(·)) = L(t, x(t), ẋ(t))tτ −1 dt 4) If the function gτ (t) is given by (2), then the
0 ODEs (4) reduce to
obtained for gτ (t) in (3).
∂L d ∂L ∂L
1.1. Proposition. The single-time Euler- − = −τ i , i = 1, ..., n.
Lagrange ODEs associated to the action (4) are ∂xi dt ∂ ẋi ∂ ẋ
1 1 ∂2L
L : R × R → R, L(x, ẋ) = − ẋ2 − axẋ − x2 gij (t, x, ẋ) = (t, x, ẋ)
2 2 ∂ ẋi ∂ ẋj
ISSN: 1109-2769 Issue 1, Volume 7, January 2008
20
WSEAS TRANSACTIONS on MATHEMATICS Constantin Udriste, Dumitru Opris
introduce the jet bundle of order one J 1 (T, M ) and Remarks. 1) A particular weight Gτ (t) can
a local chart (t, x, xα ) on it defined by a local chart be obtained taking a Riemannian diagonal mani-
t = (tα ), α = 1, ..., m, (”multitime”) on the man- fold (T, hτ α (tα )) instead the Euclidean manifold
ifold T , a local chart x = (xi ), i = 1, ..., n, on (T, δαβ ). In this case the Lagrangian is L =
∂xi
p
L(t, x(t),p xγ (t)) det (hτ α (tα )) and the weight is
the manifold M and a local chart xiα = , i =
∂tα Gτ (t) = det (hτ α (tα )).
1, ..., n; , α = 1, ..., m, on the vertical fibre. 2) If we have in mind only the Lagrangian density
Any C ∞ real function L = L(t, x(t), xα (t)) de- g 00α (tα ) ∂L
fined on J 1 (R, M ) is called Lagrangian density of en- L, then the term Fi = τ0 α in Euler-Lagrange
gτ α (t ) ∂xiα
ergy. The multi-time Stieltjes action is defined via a PDEs (9) stands for the external forces.
multiple Stieljes integral of L with respect to the func- Examples. 1) If
tions gτ α (tα ), α = 1, ..., m in the sense of functional
(τ α )rα − (τ α − tα )rα
gτ α (tα ) = , 0 < rα ≤ 1,
Z
Iτ (x(·)) = L(t, x(t), xα (t))dgτ 1 (t1 )...dgτ m (tm ) Γ(1 + rα )
Ω0τ
gτ00α (tα ) 1 − rα
or, particularly, as multitime Riemann action then 0 α
= α and the PDEs (9) are writ-
gτ α (t ) τ − tα
Z ten as multitime Euler-Lagrange PDEs with fractional
Iτ (x(·)) = L(t, x(t), xα (t))Gτ (t)dt1 ...dtm , forces
Ω0τ
m Q 0 α ∂L d ∂L 1 − rα ∂L
where Gτ (t) = α=1 gτ α (t ). We define i
− α i = α .
the multitime action of the Lagrangian density ∂x dt ∂xα τ − tα ∂xiα
L(t, x(t), xα (t)) with respect to the weight Gτ (t) by 2) If gτ α (tα ) = tα , the PDEs (9) are written as
Z the classical multitime Euler-Lagrange PDEs.
L(t, x(t), xα (t))Gτ (t)dt1 ...dtm .
α α
Iτ (x(·)) = e−τ t
Ω0τ 3) If gτ α (tα ) = − , the PDEs (9) are writ-
(8) τα
ten as Euler-Lagrange PDEs from economics
The function
∂L d ∂L ∂L
L(t, x(t), xγ (t)) = L(t, x(t), xγ (t))Gτ (t) − = −τ α i ;
∂xi dtα ∂xiα ∂xα
is called Lagrangian. tτ
α
4) If gτ α
α (t )
= α , the PDEs (9) are written as
4.1. Proposition. The multitime Euler-Lagrange
τ
PDEs associated to the action (8) are Euler-Lagrange PDEs from fractal theory of solids
∂L d ∂L gτ00α (tα ) ∂L ∂L d ∂L α ∂L
− = (9) − α i = tτ −1 i .
∂xi dtα ∂xiα gτ0 α (tα ) ∂xiα ∂xi dt ∂xα ∂xα
i = 1, ..., n; α = 1, ..., m, In order to introduce the multitime fractional
d ∂ ∂ ∂ functional like a path independent curvilinear integral,
where the symbol α
= α + xiα i + xiαβ i we start with a generic Lagrangian density of energy
dt ∂t ∂x ∂xβ
L and we build the total derivative
stands for the total derivative.
Proof. Since ∂L
Lβ (t, x(t), xα (t)) = (t, x(t), xα (t))+
∂tβ
L(t, x(t), xα (t)) = L(t, x(t), xα (t))Gτ (t)
∂L ∂xi ∂L ∂xiλ
(t, x(t), xα (t)) (t)+ i (t, x(t), xα (t)) (t).
and ∂xi ∂tβ ∂xλ ∂tβ
∂Gτ gτ00α (t)
(t) = Gτ (t), For such type of functions we define the curvilinear
∂tα gτ0 α (t)
Stieltjes functional
the classical Euler-Lagrange PDEs Z
∂L d ∂L Jτ (x(·)) = Lβ (t, x(t), xα (t))dgτ β (tβ ), (10)
− α i =0 Γ0τ
∂x i dt ∂xα
where Γ0,τ is an arbitrary piecewise C 1 curve joining
can be written as in Proposition. the points 0 and τ in Ω0τ ⊂ R+ m.
2 2
4.2 Proposition [20], [22]. 1) If x∗ (·) is an ex- e−τ t
tremal of the Lagrangian density of energy L, then and gτ 2 (t2 ) = − 2 , we find
τ
x∗ (·) is an extremal of dL.
2) If x∗ (·) is an optimum point of the functional r1 − 1
sin x − x12 = − x1 − τ 2 x2 .
Jτ (x(·)), then x∗ (·) is the solution of the multitime τ 1 − t1
Euler-Lagrange PDEs 2) (Degenerate Lagrangian). The degenerate
two-time Lagrangian
∂Lβ d ∂Lβ gτ00α (tα ) ∂Lβ
− = aβi + , (11)
∂xi dtα ∂xiα gτ0 α (tα ) ∂xiα L : J 1 (R2 , R3 ) → R,
aβi = const, i = 1, .., n; α = 1, ..., m. 1
L(t1 , t2 , x) = −(x2 )2 − (x3 )2 + x2 x11 + x3 x12
2
Commentary. 1) The fractional multitime action −x1 x21 − x1 x32 gτ0 1 (t1 )gτ0 2 (t2 )
can be represented as multiple integral or as curvilin-
ear integral. For this purpose it is enough to replace produces an Euler-Lagrange system of order one
the volume element dt1 ...dtm by dgτ 1 (t1 )...dgτ m (tm )
or the linear element (dtβ ) by (dgτ β (tβ )). 1 gτ001 (t1 ) 2 1 gτ002 (t2 ) 3
−x21 − x32 = x + x
2) The multitime dynamics with fractional action 2 gτ0 1 (t1 ) 2 gτ0 2 (t2 )
is suitable for the differential geometry of problems in 1 gτ001 (t1 ) 1
Continuous Mechanics including fractal theory. Par- −x2 + x11 = − x
ticularly, it describes qualitative properties of m-flows 2 gτ0 1 (t1 )
and their associated geometric dynamics [13]-[23]. 1 gτ001 (t1 ) 1
3) A fractional multi-time action lead to the Euler- −x3 + x12 = − x .
2 gτ0 1 (t1 )
Lagrange PDEs with external forces which are proper
for the system. 3) (Modified hyperbolic PDE). The two-time
4) Let us point out some criteria to select the Lagrangian
functions gτ β (tβ ). For example, if t1 represents
the time, then it is suitable to take gτ 1 (t1 ) = L : J 1 (R2 , R) → R,
(τ 1 )r1 − (τ 1 − t1 )r1
; if t2 represents the dilatation, 1 2
Γ(1 + r1 ) L(t1 , t2 , x) = ekt (x1 )2 ω 2 − (x2 )2 − 2kxx2
2
then gτ 2 (t2 ) = t2 ; if t3 represents the discounting,
e−τ t
3 3
−k 2 x2 gτ0 1 (t1 )gτ0 2 (t2 )
then gτ 3 (t3 ) = − 3 ; if t4 represents the fractal-
τ 4 defines the hyperbolic Euler-Lagrange PDE
4 (t4 )τ
ization, then gτ 4 (t ) = .
τ4 gτ001 (t1 ) 2 1
5) The results from [13]-[24] can be reformulated −x11 ω 2 + x22 + kx2 = ω x
gτ0 1 (t1 )
for the fractional multi-time actions.
Applications and Examples. We start from ex- 1 gτ002 (t2 )
amples in continuous mechanics [9], modified in the − (x2 + kx).
previous sense. 2 gτ0 2 (t2 )
1) (Modified sine-Gordon PDE). The two-time Taking successively
Lagrangian
L : J 1 (R2 , R) → R (τ α )rα − (τ α − tα )rα
gτ α (tα ) = tα , gτ α (tα ) = ,
Γ(1 + rα )
1
L(t1 , t2 , x) = ( x1 x2 − cos x)gτ0 1 (t1 )gτ0 2 (t2 )
2 e−τ t
α 2
The classical Hamilton PDEs can be extended to 6.2. Proposition. The multitime Euler-Lagrange
PDEs that contains the Jacobian matrix of the Legen- PDEs (9) are equivalent to the multitime Hamilton
dre transformation. PDEs
6.1. Proposition. Let x(·) be a solution of the ∂xi ∂H
(t) = β (t, x(t), p(t))
multitime Euler-Lagrange PDEs and define p(·) = ∂tβ ∂pi
(pαi (·)) as above. Then the pair (x(·), p(·)) is a solu-
∂pαi ∂H
tion respectively for the generalized multitime Hamil- α
(t) = − i (t, x(t), p(t)) + Fi (t, p(t)), (14)
ton PDEs ∂t ∂x
g 00α (t) α
∂xi ∂Hβα Fi = τ0 p .
δγα (t) = (t, x(t), p(t)) (12) gτ α (t) i
∂tβ ∂pγi
1 α λ ∂xjλ
+ λ α
δ p (t) − δβ pj (t) (t, x(t), p(t)), 7 Geometry associated to multitime
m β j ∂pγi
Euler-Lagrange derivative
1 α ∂pγi ∂Hβα Dynamical connection and semi-spray. We use the
δβ γ (t) = − (t, x(t), p(t));
m ∂t ∂xi jet bundle of order one J 1 (T, M ) and a local chart
(t, x, xα ) defined by a local chart t = (tα ), α =
∂xi ∂Tβα
δγα (t) = (t, x(t), p(t)) (13) 1, ..., m, on the manifold T , a local chart x =
∂tβ ∂pγi (xi ), i = 1, ..., n, on the manifold M and a! local
∂xj ∂xi
chart for partial velocities xα = (xiα ) = . Ex-
+ δβα pλj (t) − δβλ pαj (t) λ
(t, x(t), p(t)), ∂tα
∂pγi
plicitly, the system of local coordinates is (tα , xi , xiα ).
∂pγi ∂Tβα The manifold J 1 (T, M ) is endowed with the follow-
δβα (t) = − (t, x(t), p(t)). ing natural structures:
∂tγ ∂xi
1) the total derivative operator
Proof: Let us justify the PDEs (12). We find ∂ ∂
dα = α
+ xiα i ;
∂t ∂x
∂ 1 ∂
H α (t, x, p) = − δβα i L(t, x, xγ (t, x, p)). 2) the contact 1-forms θi = dxi − xiα dtα ;
∂xi β m ∂x
3) the total derivative 1-form operator
∂L
Now pαi (t) = (t, x(t), xγ (t)) if and only if θ1 = dα ⊗ dtα ;
∂xiα
∂xi ∂
(t) = xα (t, x(t), p(t)). Therefore the Euler- 4) the vector-valued contact form θ2 = ⊗ θi ;
∂tα ∂xi
Lagrange PDEs imply the multitime Hamilton PDEs 5) the vertical endomorphism field
in the second place. ∂
Now we compute the partial derivatives J = J α ⊗ dα , J α = ⊗ θi ,
∂xiα
∂Hβα ∂xjβ 1 α ∂L ∂xjλ ∂
α i α
γ = δγ xβ + pj γ − δ , where { } is a basis of vertical distribution V (ver-
∂pi ∂pi m β ∂xjλ ∂pγi ∂xiα
tical vector fields).
A C ∞ vector-valued 1-form H on J 1 (T, M ) is
∂xjλ
!
which contains the Jacobian matrix of the called dynamical connection on J 1 (T, M ) if it satis-
∂pγi fies the conditions
Legendre transformation. On the other hand, pαi (t) =
∂L θ1 ◦ H = 0, θ2 ◦ H = θ2 , H|V = −id|V .
(x(t), xγ (t)), implies xα (t) = xα (x(t), p(t)).
∂xiα 7.1. Proposition. The local expression of the
That is why, we get the multitime Hamilton PDEs in dynamical connection H with respect to the chart
the first place. (tα , xi , xiα ) is
Remark. After our knowledge, here is the first
time when the Jacobian matrix of the Legendre trans- ∂ ∂
H = (−xiα i
+ Hαβ ) ⊗ dtα
formation is involved in the Hamilton PDEs. ∂xi ∂xiβ
ISSN: 1109-2769 Issue 1, Volume 7, January 2008
25
WSEAS TRANSACTIONS on MATHEMATICS Constantin Udriste, Dumitru Opris
∂ j ∂ ∂ extends to a semi-spray
+( i
+ Hiα j ) ⊗ dxi − i ⊗ dxiβ .
∂x ∂xα ∂xβ
∂ ∂ ∂
Γα = + xiα i + Fαβ
i
, (15)
7.2. Proposition. 1) The rank of the matrix asso- ∂tα ∂x ∂xiβ
ciated to the dynamical connection is (m + 1)n.
where
2) The dynamical connection defines an
1
f (3, −1)-structure.
i
Fαβ = Aij
αγ φγjβ − δβγ Bj
m
As any f (3, −1)-structure, the dynamical con-
nection determines the projectors and
φβjβ = 0, Aij γ ij γ
αγ φjβ = Aβγ φjα .
` = H ◦ H = H 2 , m = −H 2 + I
Proof. To prove this statement we use two ingre-
having the following properties: dients: (1) an anti-trace PDEs system
`2 = `, m2 = m, ` ◦ m = m ◦ ` = 0, ` + m = I 1 γ
Aγα j
ij (t, x(t), xσ (t))xαβ (t) + δ Bi (t, x(t), xσ (t))
∂ ∂ ∂ m β
j j
`( ) = −xiα i − (xiα Hiβ + Hαβ ) j
∂t α ∂x ∂xβ = φγiβ (t, x(t), xσ (t)),
gτ00σ (tσ ) ∂L
!
∂Aµλ 1 α ∂L
1 ∂B
Hαβ = Aγβ Aµλ B − xα . − δγ − = 0.
m ∂xγ ∂xγ m ∂xi gτ0 σ (tσ ) ∂xiσ
Example. Let us take the PDE If the Lagrangian density of energy L is nondegener-
αβ ij
ate, then the matrix (gij ) has an inverse (gαβ ). There-
ω 2 x11 − x22 − kx2 + h1 ω 2 x1 − h2 (x2 + kx) = 0, fore a semi-spray associated to the Euler-Lagrange
PDEs is characterized by the functions
where ω, k are constants and hi = hi (t1 , t2 ), i = !
1, 2. In this case i ij 1 ∂L gτ00γ (tγ ) ∂L
Fαβ = gα δ −
11 2 12 21 22 m β ∂xj gτ0 γ (tγ ) ∂xjγ
A = ω ,A =A = 0, A = −1
!
1 ∂2L ∂2L
A11 = 2 , A12 = A21 = 0, A22 = −1. − xk − .
ω ∂xj ∂x k β ∂xj ∂tβ
It appears the semi-spray Automatically, the formulas (15) produce the compo-
1 nents of the associated dynamical connection.
F11 = − (h1 ω 2 x1 − (k + h2 )x2 − kh2 x), Poincaré-Cartan form. Let Γα , α = 1, ..., m,
2ω 2
be a semi-spray on J 1 (T, M ). The semi-spray is
F12 = F21 = 0, F22 = −ω 2 F11 called compatible to a Lagrangian
and the dynamical connection L(t, x(t), xα (t)) = L(t, x(t), xα (t))Gτ (t)
h1 k + h2 if it satisfies the multitime PDEs
H1 = − , H2 = −
2 2
∂L ∂L gτ00α (tα ) ∂L
h1 x1 (k + h2 )x2 Γα − + = 0. (17)
H11 = − , H12 = H21 = 0, H22 = − . ∂xiα ∂xi gτ0 α (tα ) ∂xiα
2 2
Now, let If the semi-spray is given by the formulas (15), then
the condition (17) leads to the PDEs
L(t, x(t), xα (t)) = L(t, x(t), xα (t))Gτ (t) αβ j
gij Fαβ + Bi = 0,
be a multitime Lagrangian. Since the d-tensor field
αβ ∂2L where
gij = is the dominant coefficient for a ge- j j αβ ∂2L
∂xiα ∂xjβ Fαβ = Fβα , gij = ,
∂xiα ∂xjβ
ometrical theory, we writte the Euler-Lagrange PDEs
of L in the form ∂2L k ∂2L ∂L gτ00γ (tγ ) ∂L
Bi = x + − + .
d ∂L ∂L gτ00α (tα ) ∂L ∂xi ∂xk j
∂t ∂x ∂xi gτ0 γ (tγ ) ∂xiγ
− + =
dtα ∂xiα ∂xi gτ0 α (tα ) ∂xiα An arbitrary dynamical connection H on
J 1 (T, M ) determines the dual bases
∂2L ∂2L j
xjαβ + x
∂xiα ∂xjβ ∂xiα ∂xj α ∂ ∂ ∂
Γα = + xiα i + Fαβ
i
∂tα ∂x ∂xiβ
∂2L ∂L gτ00α (tα ) ∂L
+ − + = 0.
∂xiα ∂tα ∂xi gτ0 α (tα ) ∂xiα ∂ 1 ∂ ∂
Hi = + Hj , Viα =
αβ j ∂xi 2 iα ∂xjα ∂xiα
This system can be identified directly to + gij xαβ
Bi = 0 and we can apply the previous theory. But, 1 i j
dtα , θi = dxi −xiα dtα , ψαi = dxiα − Hjα i
θ −Fαβ dtβ .
to show that the previous way is not unique, we prefer 2
another extension as the anti-trace PDEs system Let ω = dt1 ∧ ... ∧ dtm and
d ∂L 1 α ∂L gτ00σ (tσ ) ∂L
− δ − = ˆ α ∧ ... ∧ dtm .
ωα = (−1)m dt1 ∧ ... ∧ dt
γ
dtγ ∂xiα m ∂xi gτ0 σ (tσ ) ∂xiσ
∂L i
∂2L ∂2L ∂2L Then the m-form θ1 = Lω + θ ∧ ωα is called the
xjβγ + i j xjγ + i γ ∂xiα
∂xiα ∂xjβ ∂xα ∂x ∂xα ∂t Poincaré-Cartan form.
ISSN: 1109-2769 Issue 1, Volume 7, January 2008
27
WSEAS TRANSACTIONS on MATHEMATICS Constantin Udriste, Dumitru Opris
∂L ∂L
7.5. Proposition. The (m + 1)-form Ω1 = dθ1 Jijα = Hi − Hj .
can be written ∂xjα ∂xiα
αβ i 1
Ω1 = (gij ψβ ∧ θj − Jij1α θi ∧ θj )ωα + Ei1 (L)θi ∧ ω, 7.8 Proposition. If the dynamical connection H
2 is associated to a semi-spray which is compatible to
where L, i.e., Ei (L) = 0, then
∂L ∂L ∂L ∂L 1
Ei1 (L) = i
−Γα i , Jij1α = Hi ( j )−Hj ( i ). αβ i
Ω1 = (gij ψβ ∧ θj − Jijα θi ∧ θj )Gτ (t)ωα
∂x ∂xα ∂xα ∂xα 2
m−1 ∂Bj ∂Bi 1 ∂φαiβ ∂φαjβ
7.6 Proposition. If the dynamical connection H
Jijα = − + − ,
is associated to a semi-spray which is compatible to 2m ∂xiα ∂xjα 2 ∂xjβ ∂xiβ
L, i.e., Ei1 (L) = 0, then
where
1 αβ i 1
Ω = (gij ψβ ∧ θ − Jij1α θi ∧ θj )ωα
j
2 ∂2L ∂2L ∂L gτ00α (tα ) ∂L
Bj = xiα + − + .
∂xi ∂xjα ∂tα ∂xjα ∂xj gτ0 α (tα ) ∂xiα
∂Bj1 ∂Bi1 1 ∂φαiβ ∂φαjβ
!
m−1
Jij1α = − + − ,
2m ∂xiα ∂xjα 2 ∂xjβ ∂xiβ
where 8 Multitime Hamilton-Poisson sys-
∂2L ∂2L ∂L tems on jet bundle
Bj1 = i
j xα + − .
i
∂x ∂xα ∂tα ∂xjα ∂xj
If (T, h) and (R, g) are Riemannian manifolds, we
shall use the adapted dual bases
If m = 1, then Ω1 = gij ψ i ∧ θj , gij =
∂L i dxi δ ∂ ∂
, ẋ = . If n = 1, then Ω1 = hαβ ψα ∧ γ
= α + Hαβ xiγ i ,
∂ ẋi ∂ ẋj dt δt α ∂t ∂xβ
∂L
θ ∧ ωβ , hαβ = .
∂xα ∂xβ δ ∂ ∂ ∂
The previous theory refers to classical Riemann i
= i
− Ghik xkα h ,
actions. Its reformulation for Stieltjes actions is obvi- δx ∂x ∂xα ∂xiα
ous. For example, the Poincaré-Cartan m-form can be (dtβ , dxj , δxjβ = dxjβ − Hβλ xγ dt + Gjhk xhβ dxk )
γ j λ
∂L
written θ = Lω̃ + i θi ∧ ω̃α , where
∂xα as frames on the jet bundle J 1 (T, M ). Then the in-
duced Riemann Sasaki-like metric on J 1 (T, M ) is
ω̃ = dgτ 1 (t1 ) ∧ ... ∧ dgτ m (tm ),
S = hαβ dtα ⊗dtβ +gij dxi ⊗dxj +hαβ gij δxiα ⊗δxjβ .
m ˆ τ α (tα )∧...∧dgτ m (tm ).
1
ω̃α = (−1) dgτ 1 (t )∧...∧ dg
gτ00α (tα ) We first notice that, on the Riemannian manifold
Since ω̃ = Gτ (t)ω, ω̃α = ωα , we can writte (J 1 (T, M ), S) there exists a globally defined 1-form
gτ0 α (tα )
d-tensor
∂L gτ00α (tα ) i ω = gij xjα dxi ⊗ dtα .
θ = Gτ (t)Lω + θ ∧ ωα .
∂xiα gτ0 α (tα ) Its exterior differential
7.7 Proposition. The (m + 1)-form Ω = dθ can Ω = dω = (−gij dxi ∧ δxjα ) ⊗ dtα
be written
1
gτ00α (tα ) is also globally defined 2-form d-tensor, and has the
αβ i
Ω = gij ψβ ∧ θj − Jijα θi ∧ θj ωα components
2 gτ0 α (tα )
−gij δαβ
!
i 0
+Ei (L)Gτ (t)θ ∧ ω, (Ωα AB ) =
where gij δαβ 0