of variations
D. Krupka and M. Krupka
Dedicated to the 70-th anniversary of Professor Constantin Udriste
Abstract. Geometric structure of global integral variational functionals on higher order tangent bundles and Grassmann fibrations are investi-gated. The theory of Lepage forms is extended to these structures. The concept of a Lepage form allows us to introduce the Euler-Lagrange dis-tribution for variational functionals, depending on velocities, in a similar way as in the calculus of variations on fibred manifolds. Integral curves of this distribution include all extremal curves of the underlying variational functional. The generators of the Euler-Lagrange distribution, defined by the Lepage forms of the first order, are found explicitly.
M.S.C. 2000: 49Q99, 58A20, 58A30, 58E99.
Key words: Variational theory; velocity bundle; Grassmann bundle; Lepage form; Euler-Lagrange distribution.
1
Introduction
Our main objective in this paper is the geometric structure of the variational theory on higher order velocity spaces and Grassmann fibrations. We consider global integral variational functionals forcurvesand 1-dimensional submanifoldsof a given manifold. A specific feature of this theory is the absence of the concept of aLagrangian, a basic element of the classical calculus of variations and the variational theory on fibred manifolds; in this paper the role of a Lagrangian is played by a differential 1-form on a velocity manifold, called aLepage form.
We introduce Lepage forms, and derive a geometric (coordinate-free) first varia-tion formula on thehigher order velocity spaces. Our definition extends properties of the Cartan and Lepage forms, used in classical mechanics and the global variational theory on fibred manifolds. Main tools we use are properties of forms on the manifolds of velocities, and independence of the variational integral on parametrization. These notions are naturally characterized via the theory of jets and higher order contact elements. We also describe the Euler-Lagrange distribution, related with a Lepage form, whose integral curves include all extremals of the variational integral. In par-ticular, we derive chart expressions for a Lepage form as well as the generators of
∗
Balkan Journal of Geometry and Its Applications, Vol.15, No.1, 2010, pp. 68-79. c
the Euler-Lagrange destribution on the first order Grassmann fibration in terms of
adapted coordinates. The reader can easily understand that the presented theory can be extended to variational functionals for submanifolds of dimension greater than 1. Throughout this paper, R is the field of real numbers. We denote by Y a fixed smooth manifold of dimension m+ 1. TrY is the manifold of velocities of order r
over Y (r-jets Jr
0ζ with source 0∈ R and target ζ(0) ∈ TrY, ) andτr,s : TrY →
TsY are the canonical jet projections. ImmTrY denotes an open submanifold of
regular velocities inTrY (r-jets of immersions), andLr is the differential group of
order r of R , consisting of regular r-jets Jr
0α∈ ImmTrR suchα(0) = 0, with the group multiplication the composition of jets. GrY is the Grassmann fibration of
order r over Y (the quotient manifold ImmTrY /Lr ); ρr,s : GrY → GsY denotes
the canonical projection of Grassmann fibrations. For simplicity, we restrict our coordinate considerations to lower order casesr= 1,2.
For the general theory of jets and contact elements the reader is referred to the papers [3, 5, 6]. The theorems, presented in this paper, namely the structure theory of Lepage forms and a new description of extremals in terms of a distribution, are an extension of the variational calculus on fibred manifolds as explained in [1, 2, 4, 7].
2
Velocities and Grassmann fibrations
Given a chart (V, ψ), ψ = (yK), on a manifold Y of dimension m+ 1, the
associ-ated charts on the manifolds of regular velocities ImmT1Y and ImmT2Y are denoted by (V1, ψ1), ψ1 = (yK,y˙K), and (V2, ψ2), ψ2 = (yK,y˙K,y¨K). Each index L
de-fines a partition of the index set, {L}, {1,2,· · ·, L−1, L+ 1,· · ·, m, m+ 1}, and thesubordinate charts on ImmT1Y and ImmT2Y, denoted by (V1,L, ψ1,L), ψ1,L =
(yL,y˙L, yσ,y˙σ), and (V2,L, ψ2,L),ψ2,L= (yL,y˙L,y¨L, yσ,y˙σ,y¨σ) ; recall that the sets
V1,LandV2,L are defined by
˙
wL6= 0.
(2.1)
The corresponding second subordinate charts are denoted by (V1,L, χ1,L), χ1,L =
(wL,w˙L, wσ, wσ
1), and (V2,L, χ2,L), χ2,L = (wL,w˙L,w¨L, wσ, wσ1, w2σ). The second subordinate charts can be introduced by the transformation equations.
Lemma 1. The transformation equations between the charts (V2,L, ψ2,L) and
(V2,L, χ2,L)are
yL=wL, y˙L= ˙wL, y¨L = ¨wL
(2.2)
yσ=wσ, y˙σ=wσ
1w˙L, y¨σ=wσ2( ˙wL)2+w1σw¨L.
LetLr be the differential groupof order r of the real line R; in the context of this
work,Lrdescribes the change of parameter in variational functionals for curves inY.
Lracts canonically onTrY to the right by composition of jets,
TrY ×Lr∋(Jr
0ζ, J0rα)→J0rζ◦J0rα=J0r(ζ◦α)∈TrY. (2.3)
Clearly, this group action restricts to the submanifold ofregularvelocities ImmTrY.
Recall that the canonical coordinates a1, a2,· · ·, ar on Lr are the functions on Lr
Lemma 2. (a)The group multiplication(J2
0α, J02β)→J02(α◦β)in the groupL2
is expressed by the equations
a1(Jr
0α◦J0rβ) =a1(J0rα)·a1(J0rβ), (2.4)
a2(J0rα◦J0rβ) =a2(J0rα)·(a1(J0rβ))2+a1(J0rα)·a2(J0rβ).
(b)The group action (2.3), restricted to the set ImmT2Y, is expressed in a
sub-ordinate chart(V2,L, χ2,L),χ2,L= (wL,w˙L,w¨L, wσ, wσ
1, w2σ), by the equations
wL(J2
0ζ◦J02α) =wL(J02ζ), (2.5)
˙
wL(J2
0ζ◦J02α) = ˙wL(J02ζ)a1(J02α), ¨
wL(J2
0ζ◦J02α) = ˙wL(J02ζ)a2(J02α) + ¨wL(J02ζ)a1(J02α)2,
wσ(J2
0ζ◦J02α) =wσ(J02ζ),
wσ1(J02ζ◦J02α) =wσ1(J02ζ),
wσ
2(J02ζ◦J02α) =wσ2(J02ζ).
Lemma 3. Suppose we have two charts on Y,(V, ψ), ψ= (yK), and(V,ψ¯),ψ¯=
(¯yK)such thatV ∩V¯ 6=¡❢, and the transformation equations
¯
yM =fM(yL, yσ), y¯ν =fν(yL, yσ).
(2.6)
Then the transformation equations between the subordinate charts(V1,L, χ1,L),χ1,L=
(wL,w˙L, wσ, wσ
1)and( ¯V1,M,χ¯1,M),χ¯1,M = ( ¯wM,w˙¯M,w¯σ,w¯1σ), are ¯
wM =fM(wL, wσ), w¯ν =fν(wL, wσ), w˙¯M = ∂fM
∂wLw˙
L+∂fM
∂wσw˙ Lwσ
1, (2.7)
¯
wν
1 =
1
∂fL
∂wL+
∂fL
∂wτw τ
1 µ
∂fν
∂wL +
∂fν
∂wσw σ
1 ¶
.
Let GrY be the Grassmann fibration of order r over Y. We denote by [Jr
0ζ] the Lr-orbit of a regular velocity Jr
0ζ, and by ρr,s : GrY → GsY the canonical projection. For every indexLthe pair ( ˜V2,L,χ˜2,L), where ˜V2,L= (ρ2,0)−1(V),χ˜2,L= ( ˜wL,w˜σ,w˜σ
1,w˜2σ), and for allJ02ζ∈V2,L ˜
wL([J2
0ζ]) =wL(J02ζ), (2.8)
˜
wσ([J2
0ζ]) =wσ(J02ζ), w˜σ1([J02ζ]) =wσ1(J02ζ), w˜σ2([J02ζ]) =w2σ(J02ζ), is a chart onG3Y.
LetIbe an open interval, containing 0∈R, and letγ:I→Y be a curve. Denote by trt0 the translationt→t−t0 of R. γdefines ther-jet prolongation
I∋t→(Trγ)(t) =J0r(γ◦tr−t)∈TrY.
(2.9)
Further on, we suppose thatγis animmersionsuch that for a chart onY,γ(I)⊂V
Lemma 4. The 2-jet prolongationT2γ is expressed by
wL◦T2γ=wLγ, w˙L◦T2γ=D(wLγ), w¨L◦T2γ=D2(wLγ),
(2.10)
wσ◦T2γ=wσγ, wσ
1 ◦T2γ=
D(wσγ)
D(wLγ),
wσ2◦T2γ= 1 (D(wLγ))2
µ
D2(wσγ)−D
2(wLγ)
D(wLγ)D(w σγ)
¶
.
Ther-jet prolongation of an immersionγ:I→Y defines a curve inGrY,
I∋t→Grγ(t) = [Trγ(t)]∈GrY,
(2.11)
called theGrassmann prolongationofγ of orderr.
We examine the behavior of the mappingT2γunderreparametrizations. LetJ be an open interval, containing the origin 0, and letµ:J→Ibe a diffeomorphism. µis defined by an equation
t=µ(s).
(2.12)
Setting for everys∈J
µs(t) = trµ(s)◦µ◦tr−s(t) =−µ(s) +µ(s+t), (2.13)
we get another diffeomorphism µs : Js →Is of open intervals, containing 0. Since
Dµs(t) =Dµ(s+t) andD2µs(t) =D2µ(s+t),µssatisfies
µs(0) = 0, Dµs(0) =Dµ(s), D2µs(0) =D2µ(s).
(2.14)
In particular, the 2-jetJ2
0µsis an element of the differential groupL2for alls, whose
canonical coordinates area1(s) = Dµ(s), a2(s) =D2µ(s). µinduces a differentiable mappings→J2
0µsof the domainJ ofµintoL2;µalso induces a diffeomorphism
ImmT2Y ∋J02ζ→J02ζ◦J02µs∈ImmT2Y,
(2.15)
defined by the canonical action of L2 on ImmT2
0Y. The mappings s → J02µs and
J2
0ζ→J02ζ◦J02µs are said to beassociatedwithµ.
A diffeomorphismµ:J →I assigns to the immersionγan immersionγ◦µ:J →
Y, and its 2-jet prolongationT2(γ◦µ).
Lemma 5. (a)The mappings→T2(γ◦µ)(s) satisfies
T2(γ◦µ)(s) =T2γ(µ(s))◦J02µs.
(2.16)
(b)The mapping s→T2(γ◦µ)(s)is expressed in the chart (V2,L, χ2,L),χ2,L =
(wL, wL
1, w2L, wσ, wσ1, wσ2)by
wL(T2(γ◦µ)(s)) =wL(T2γ(µ(s))),
(2.17)
wL
1(T2(γ◦µ)(s)) =w1L(T2γ(µ(s)))a1(J02µs),
wL
2(T2(γ◦µ)(s)) =w1L(T2γ(µ(s)))a2(J02µs) +wL2(T2γ(µ(s)))a1(J02µs)2,
wσ(T2(γ◦µ)(s)) =wσ(T2γ(µ(s))),
wσ
1(T2(γ◦µ)(s)) =w1σ(T2γ(µ(s))),
There exists a bijection between the set of diffeomorphismsα: (−a, a)→R such
equal to the identity elementJ2
0idR of the groupL2,
immediately be modified for vector fields by means of flows; we denote by Trξ the
r-jet prolongationof a vector fieldξonY.
Lemma 6. Letξ be a vector field onY, and let
3
The calculus of variations on velocity manifolds
Choose a velocityP ∈T3Y and a representativeζ ofP; then P =J3
0ζ. ζdefines a mappingT2ζof a neighbourhood of 0∈R intoT2Y and the tangent mapping at 0,
T0T2ζ :T0R→TJ2 0ζT
2Y, sending a vector ξ∈T0R to a vector of T2Y at the point
T2ζ(0) =J2
0ζ =τ3,2(J03ζ). Express ξis in the canonical basis of the 1-dimensional vector spaceT0R as ξ =ξ0·(d/dt)0 and define a vector field δalong the projection
τ3,2 by
δ(J03ζ) =T0T2ζ· µ
d dt
¶
0
.
(3.1)
The vector fieldδinduces a mappingη→hη, defined on differential 1-forms onT2Y, with values in the module of functions onT3Y by
hη(J03ζ) =η(J02ζ) ¡
δ(J03ζ) ¢
.
(3.2)
In particular, iff is a function on an open setW ⊂T2Y, then the formula δ(f) =
h(df) defines a functionδ(f) on the set (τ3,1)−1(W)⊂T3Y.
Lemma 7. Let η be a 1-form, let (V, ψ), ψ = (yK), be a chart onY, and let γ be a curve with values inV.
(a)δ has a chart expression
δ= ˙yK ∂
∂yK + ¨y K ∂
∂y˙K.
(3.3)
(b)If η is expressed asη=AkdyK+BKdy˙K, then
hη=AKy˙K+BKy¨K.
(3.4)
We callδ theformal derivative morphism; the functionδ(f) is called theformal derivativeoff. The mappinghis called thehorizontalization.
Remark 1. From the definitions (3.1) and (3.2) we easily derive the formulas
hdwL= ˙wL, hdw˙L = ¨wL, hdwσ = ˙wLw1σ, hdw1σ= ˙wLw2σ. (3.5)
LetW be an open set inY, and suppose we have a 1-formη, defined on the set (τr,0)−1(W)⊂ImmTrY. We say thatηiscontact, ifTrζ∗η= 0 for all immersionsζ, defined on an open interval in R, with values inW. The ideal of the exterior algebra of differential forms on the set (τr,0)−1(W), locally generated by contact 1-forms, is called thecontact ideal, and is denoted by Ω1cW. By a contactk-form we mean any
k-form, belonging to the contact ideal.
Lemma 8. LetW be an open set inY, letη be a1-form on(τ2,0)−1(W), and let (V, ψ), ψ= (yK), be an chart onY such thatV ⊂W. Then the following conditions are equivalent:
(a)η is a contact form.
(b)For every subordinate chart(V2,L, ψ2,L), ψ2,L= (yL,y˙L,y¨L, yσ,y˙σ,y¨σ),
η= ˙ALη˙L+Aσησ+ ˙Aση˙σ,
where
˙
ηL=dy˙L−y¨L
˙
yLdy
L, ησ =dyσ−y˙σ
˙
yLdy
L, η˙σ=dy˙σ−y¨σ
˙
yLdy L.
(3.7)
(c)For every subordinate chart(V2,L, χ2,L),χ2,L= (wL, wL
1, w2L, wσ, wσ1, wσ2),
η=BL1ωL
1 +Bσ0ωσ0 +Bσ1ω1σ, (3.8)
where
ω1L=dw1L−
wL
2
wL
1
dwL, ω0σ=dwσ−wσ1dwL, ωσ1 =dwσ1 −w2σdwL. (3.9)
(d)η belongs to the kernel of the horizontalization h, i.e.,hη= 0.
Clearly, the forms (3.7), and the forms (3.8) are linearly independent.
Suppose we have a 1-form η, defined on ImmT1Y. Let I be an open interval, and let γ : I → Y be an immersion. Any compact subintervalK of I defines the
variational integral, associated withη,
ηK(γ) =
Z
K
(T1γ)∗η. (3.10)
The mappingηK is theintegral variational functional, associated with theη.
The function hηis the Lagrange function, associated withη. The following gives us a description of variational functionals in terms of Lagrange functions.
Lemma 9. Let η be a 1-form on T1Y, and let γ : I → Y be an immersion,
defined on an open intervalI⊂R. Then
Trγ∗η= (hη◦Tr+1γ)·dt. (3.11)
Lemma 9 says that the Lagrange functionLη, associated withη, is given by
Lη(J0r+1ζ) =hη◦Tr+1(ζ◦trt)(t).
(3.12)
LetC2
KY denote the set of curves inY of classC2, defined on a compact interval
K⊂R. We have for every isomorphismαofY and every curveγ∈C2
KY
ηK(αγ) =
Z
K
(T1(αγ))∗η. (3.13)
But by definitions,T1(αγ) =T1α◦T1γ, so (3.13) reduces to
ηK(αγ) =
Z
K
(T1α◦T1γ)∗η=
Z
K
T1γ∗T1α∗η. (3.14)
Consequently, the variational functionalC2
KY ∋γ→ηK(αγ)∈R (3.13) satisfies
ηK(αγ) = (T1α∗η)K(γ),
(3.15)
This property of variational functionals can be transfered to vector fields. Letξ
be a vector field on Y and αξ
s its flow. Then for all sufficiently smalls, ηK(αξsγ) =
((T1αξ
s)∗η)K(γ). Differentiating, we have
µdη
K(αξsγ)
ds
¶
0 =
Z
K
T1γ∗∂
T1 ξη,
(3.16)
where ∂T1
ξη is the Lie derivative of the form η by the vector field T1ξ, the 1-jet
prolongation ofξ. The mapping
CK2Y ∋γ→(∂T1
ξη)K(γ) =
Z
K
(T1γ)∗∂
T1 ξη∈R
(3.17)
is called thefirst variationof the variational functional ηK by the vector fieldξ.
Note that the Lie derivative∂T1
ξηunder the integral sign in (3.17) can be
decom-posed as ∂T1
ξη = iT1
ξdη+diT1
ξη. The form η is said to be a Lepage form, if the
2-formdη belongs to the contact ideal Ω1
cY.
Theorem 1. (The structure of Lepage forms)The following conditions are equivalent:
(a)η is a Lepage form on ImmT1Y.
(b)η has in any subordinate chart(V1,L, ψ1,L)an expression
η=PLdyL+
∂PL
∂y˙σy˙
Lησ+dF L,
(3.18)
whereFL andPL are function onV1,L, andPL satisfies
∂PL
∂y˙Ly˙
L+∂PL
∂y˙σy˙ σ= 0.
(3.19)
(c)η has in a subordinate chart(V1,L, χ1,L)an expression
η=PLdwL+
∂PL
∂wσ
1
ωσ+dF,
(3.20)
wherePL andF are functions onV1,L such that
∂PL
∂wL
1 = 0.
(3.21)
To demonstrate basic ideas of the proof, we show that (b) follows from (a). Con-sider the chart expression
η=ALdyL+Aσησ+BLdy˙L+Bσdy˙σ,
(3.22)
where
ησ =dyσ− y˙σ
˙
yLdy L
(Lemma 8). Now
The conditions thatdη be generated by the contact formsησ imply
∂BL
Integrating these conditions we get
BL=
for a functionF. Then, however,
dF = ∂F
From this expression we have
dη= ∂A˜L
Suppose we have a Lepage form η on the manifold of velocities ImmT1Y. We wish to describe a distribution ∆η on ImmT1Y, defined by differential 1-formsiΞdη, where Ξ runs through vector fields on ImmT1Y; ∆
ηis theEuler-Lagrange distribution
associated withη.
We give an explicit characterization of the Euler-Lagrange distribution of a Lepage formη in terms of the second subordinate charts. We know that
η=LLdwL+
Theorem 2. (The Euler-Lagrange distribution) In a second subordinate chart, the Euler-Lagrange distribution∆η is generated by the 1-forms
To prove the theorem, we contract the formdη with a vector field
Ξ = ΞL ∂
∂wL + Ξ ν ∂
∂wν + Ξ L
1
∂ ∂wL
1 + Ξν
1
∂ ∂wν
1
,
(3.33)
and obtain the generators of the distribution by calculating the coefficients in the chart expression for the formiΞdη at the components ΞL, Ξσ, and Ξν1.
Theorem 2 describes important special cases. In particular, if the matrix
∂2L
L
∂wν
1∂w1σ (3.34)
is non-singular, then each integral curve of the Euler-Lagrange distribution is holo-nomic, i.e., is the prolongation of a curve in the manifoldY. In this case the generators of the Euler-Lagrange distribution reduce to
ωσ,
µ
−∂LL
∂wσ +
∂2L
L
∂wL∂wσ
1
+ ∂
2L
L
∂wν∂wσ
1
wν
1 ¶
dwL+ ∂2LL
∂wν
1∂wσ1
dwν
1. (3.35)
4
Parameter invariance
Suppose we have a 1-formηon ImmT1Y. Consider the variational integral (3.10). We are interested in the case when the numberηK(γ) is independent of parametrization
of the setγ(I). This is characterized by the following theorems.
Theorem 3. Let η be a 1-form on T1Y, let γ : I → Y be an immersion, J
an open interval, and µ : J → I a diffeomorphism. The following conditions are equivalent:
(a)For any two compact intervalsL⊂J andK⊂I such that µ(L) =K,
ηK(γ) =ηL(γ◦µ).
(4.1)
(b)η satisfies
(T1γ)∗η= (µ−1)∗T1(γ◦µ)∗η. (4.2)
Condition (4.2) is called the invariance condition; we say that η and γ satisfy the invariance condition, if (4.2) holds for all diffeomorphisms µ. We say thatη is
parameter-invariant, if (4.2) holds for allγ andµ.
Consider the variational functional (3.10). The form (T1γ)∗η has at every point
t0∈I an expression
(T1γ)∗η(t0) =L ◦T2γ(t0)·dt(t0). (4.3)
We can now give a version of the invariance condition in terms of the Lagrange functionL.
Lemma 10. Let the immersionγ and the diffeomorphism µbe given. Then the following two conditions are equivalent:
(2)For alls∈K,
L¡
J02(γ◦tr−µ(s))
¢
·Dµs(0) =L¡J02(γ◦tr−µ(s))◦J02µs¢.
(4.4)
The following is a criterion of invariance of the formηunder changes of parametriza-tion; the criterion says that the Lagrange functionL, associated with η, should be
L2-equivariant.
Theorem 4. η satisfies the invariance condition if and only if
L¡
J02γ ¢
·Dα(0) =L¡
J02γ◦J02α ¢ (4.5)
for allJ2
0α∈L2 .
Remark 2. According to Lemma 2, the group action ofL2 on ImmT2Y is in a chart (V, ψ), ψ= (yK), given by the equations ¯yK=yK,y˙¯K =ay˙K,y¨¯K =ay¨K+by˙K,
where a and b are the canonical coordinates on L2. Hence condition (3.32) can be expressed asL(yK, ay˙K, ay¨K+by˙K) =aL(yK,y˙K,y¨K).
Remark 3. Condition (4.5) can also be expressed in a subordinate chart (V2,L, χ2,L), χ2,L = (wL,w˙L,w¨L, wσ, wσ
1, w2σ). Since the group action of L2 in this chart is given by the equations ¯wL = wL,w˙¯L = aw˙L,w¯¨L = bw˙L +a2w¨L,w¯σ =
wσ,w¯σ
1 =wσ1,w¯σ2 =w2σ, whereaandb are the canonical coordinates onL2, we have
L(wL, aw˙L, bw˙L+a2w¨L, wσ, wσ
1, w2σ) =aL(wL,w˙L,w¨L, wσ, wσ1, wσ2).
We are now in a position to give a complete description of Lepage forms on ImmT1Y that satisfy the invariance condition.
Theorem 5. Let η be a 1-form on ImmT1Y. The following two conditions are
equivalent:
(a)η is a Lepage form, and satisfies the invariance condition.
(b) In every subordinate chart (V1,L, χ1,L), χ1,L = (wL,w˙L, wσ, wσ
1) has an
ex-pression
η=PLdwL+
∂PL
∂wσ
1
ωσ+dF L,
(4.6)
wherePL andFL are functions on the setV1,L such that
∂PL
∂w˙L = 0,
∂FL
∂w˙L = 0.
(4.7)
Express the Lepage formη as in Theorem 1, by
η=PLdwL+
∂PL
∂wσ
1
ωσ+dF,
(4.8)
and compute the corresponding Lagrange function L = hη. We have, using the formulashdwL= ˙wL, hdw˙L= ¨wL, hdwσ= ˙wLwσ
1, andhdw1σ= ˙wLw2σ (2.10),
L= µ
PL+
∂F ∂wL
¶ ˙
wL+ ∂F
∂w˙Lw¨
L+ ∂F
∂wσw˙ Lwσ
1 +
∂F ∂wσ
1 ˙
wLwσ
ButL(wL, aw˙L, bw˙L+a2w¨L, wσ, wσ
1, w2σ) =aL(wL,w˙L,w¨L, wσ, wσ1, wσ2) for alla, b∈ R,a6= 0, sinceηsatisfies the invariance condition (Theorem 3), andPL=PL(wL, wσ, wσ1) sinceη is Lepage; then (a) implies∂F/∂w˙L= 0.
Remark 4. From Theorem 5 we conclude that Theorem 2 remains valid for the Euler-Lagrange distribution of parameter-invariant variational problems.
Acknowledgments. The research of the first author was supported by grants 201/09/0981 (Czech Science Foundation) and MEB 080808 (Kontakt).
References
[1] J. Brajercik,GLn(R)-invariant variational principles on frame bundles, Balkan
J. Geom. Appl. 13, 1 (2008), 1-20.
[2] M. Crampin and D.J. Saunders, The Hilbert-Caratheodory form for parametric multiple integral problems in the calculus of variations, Acta Appl. Math. 76 (2003), 37-55.
[3] D.R. Grigore and D. Krupka,Invariants of velocities and higher order Grassmann bundles, J. Geom. Phys 24 (1998), 244-264.
[4] D. Krupka,Variational sequences in mechanics, Calc. Var. 5 (1997) 557-583. [5] D. Krupka and M. Krupka,Jets and contact elements, in: Proc. of the Seminar
on Differential Geometry, D. Krupka, Ed., Mathematical Publications Vol. 2, Silesian Univ. Opava, Czech Republic, 2000, 39-85.
[6] D. Krupka and Z. Urban, Differential invariants of velocities and higher order Grassmann bundles, in: Diff. Geom. Appl., Proc. Conf., in Honour of L. Euler, O. Kowalski, D. Krupka, O. Krupkova and J. Slovak, Editors, Olomouc, August 2007, World Scientific, 2008, 463-473.
[7] O. Krupkova, The Geometry of Ordinary Variational Equations, Lecture Notes in Mathematics 1678, Springer, 1997.
Authors’ addresses:
D. Krupka
Institute of Theoretical Physics and Astrophysics, Faculty of Science, Masaryk University,
Kotlarska 2, 638 00 Brno, Czech Republic;
Department of Mathematics, La Trobe University, Melbourne, Bundoora, Victoria 3086, Australia.
M. Krupka
Department of Computer Science, Faculty of Science, Palacky University,