• 検索結果がありません。

on frame bundles

N/A
N/A
Protected

Academic year: 2022

シェア "on frame bundles"

Copied!
9
0
0

読み込み中.... (全文を見る)

全文

(1)

on frame bundles

J. Brajerˇc´ık

Abstract. Variational principles on frame bundles, given by the first and the second order Lagrangians invariant with respect to the struc- ture group, are considered. Noether’s currents, associated with the corre- sponding Lepage equivalents, are obtained. It is shown that for the first and the second order invariant variational problems, the system of the Euler-Lagrange equations for a frame field are equivalent with the lower order system of equations.

M.S.C. 2000: 49Q99, 49S05, 58A20, 58E30, 58J99.

Key words: Frame, variational principle, Lagrangian, Euler-Lagrange equations, invariant transformation, Noether’s current.

1 Introduction

LetF X be the frame bundle over ann-dimensional manifoldX, and letJrF Xbe the r-jet prolongation of F X. We shall consider JrF X with the canonical prolongation of the right action of the general linear groupGln(R) onF X. For foundations of the variational theory in fibered space we refer to [5], [7], [8], [10], [14], and the notions related to the frame bundles and invariance can be found in [6], [9], [11], [12], [15]. In this paper we study the consequences ofGln(R)-invariance for variational problems onJ1F X andJ2F X. In particular, we discuss the corresponding Noether’s currents.

The generators of invariant transformations are the fundamental vector fields of the Gln(R)-action. Then the Noether’s theorem gives us a conservation law for each one ofn2linearly independent fundamental vector fields. Our main object is to study how the Noether’s currents can be used to simplify the Euler-Lagrange equations for a frame field. We show that in case of first order invariant Lagrangian, the system of n2second order Euler-Lagrange equations is equivalent with the system of the same number of first order equations. Analogously, for the second order Lagrangian, the system of fourth order Euler-Lagrange equations is equivalent to the system of third order equations coming from the corresponding Noether’s currents.

For variational problems on principal fiber bundles there are several different con- cepts of invariance. Castrill´on, Garc´ıa, Ratiu and Shkoller [3], [4] consider invariance of

Balkan Journal of Geometry and Its Applications, Vol.13, No.1, 2008, pp. 11-19.

c

°Balkan Society of Geometers, Geometry Balkan Press 2008.

(2)

the first order Lagrangians on principal fiber bundleP, which determine constrained variational problems on the bundleC(P) of connections ofP. Mu˜noz and Rosado [13]

study first order variational problems, invariant under diffeomorphisms of the base manifold (first ordernaturalvariational problems in the sense of Krupka [7]).

2 Invariant Lagrange structures

In this section we recall basic notions of the theory of invariant Lagrangians, and introduce our notation. For a more complete discussion we refer to [2].

IfY is a fibered manifold over ann-dimensional manifoldX, of dimensionn+m, we denote byJrY ther-jet prolongationofY, andπr,s:JrY →JsY,πr:JrY →X are the canonical jet projections. Ther-jetof a section γ of Y at a point x∈X, is denotedJxrγ; andx→Jrγ(x) =Jxrγ is ther-jet prolongationofγ. Anyfibered chart (V, ψ),ψ= (xi, yσ), onY, where 1≤i≤n, 1≤σ≤m, induces theassociated charts onXand onJrY, (U, ϕ),ϕ= (xi), and (Vr, ψr),ψr= (xi, yσ, yjσ1, yσj1j2, . . . , yjσ1j2...jr), respectively; hereVr= (πr,0)−1(V), andU =π(V). Recall that theformal derivative operatoris defined by

di=

∂xi +yσi

∂yσ +yiσ1i

∂yσi1 +. . .+yσi1i2...iri

∂yσi1i2...ir.

For any open set W Y, Ωr0W denotes the ring of smooth functions on Wr. The Ωr0W-module of differentialq-forms onWr is denoted by ΩrqW, and the exterior algebra of forms on Wr is denoted by ΩrW. The module of πr,0-horizontal (πr- horizontal)q-forms is denoted by Ωrq,YW (Ωrq,XW, respectively).

Thehorizontalization is the exterior algebra morphismh: ΩrW r+1W, de- fined, in any fibered chart (V, ψ),ψ= (xi, yσ), by

hf=f◦πr+1,r, hdxi=dxi, hdyjσ1j2...jp =yσj1j2...jpkdxk,

where f : Wr R is a function, and 0 p r. A form η rkW is contact, if = 0. For any fibered chart (V, ψ),ψ= (xi, yσ), the 1-forms

ωσj1j2...jp=dyjσ1j2...jp−yσj1j2...jpkdxk,

where 0≤p≤r−1, are examples of contact 1-forms. η isπr-horizontalif and only if (πr+1,r)η=hη.

A Lagrangian (of order r) for Y is any πr-horizontal n-form on some Wr. A differential form ρ snW, where n = dimX, is called a Lepage form, if p1 is πs+1,0-horizontal, i.e.p1dρ∈s+1n+1,YW. A Lepage formρis aLepage equivalent of a Lagrangianλ∈rn,XW, if=λ(possibly up to a jet projection).

In a fibered chart (V, ψ),ψ= (xi, yσ), denote

ω0=dx1∧dx2∧. . .∧dxn, ωk=i∂/∂xkω0.

In this fibered chart, a Lagrangian, defined onVr= (πr,0)−1(V), has an expression

(2.1) λ=0,

(3)

where L : Vr R is the Lagrange function associated with λ and (V, ψ). A pair (Y, λ), consisting of a fibered manifoldY and a Lagrangianλof orderrforY is called aLagrange structure(of orderr).

For our purpose we give the following examples of Lepage equivalents.

(1) Every first order Lagrangian λ 1n,XW has a unique Lepage equivalent Θλ1n,YW whose order of contactness is1. Ifλis expressed by (2.1), then

(2.2) Θλ=0+ ∂L

∂yiσωσ∧ωi.

Θλ is thePoincar´e-Cartan equivalentofλ, or thePoincar´e-Cartan form.

(2) Formula

(2.3) Θλ=0+ Ã

∂L

∂yσi −dp ∂L

∂yσpi

!

ωσ∧ωi+ ∂L

∂yσjiωσj ∧ωi

generalizes the Poincar´e-Cartan form tosecond orderLagrangiansλ∈2n,XW. Ifρis a Lepage equivalent of a Lagrangianλ∈rn,XW,λ=0, then by a direct calculationp1=Eσ(L)ωσ∧ω0, where

Eσ(L) = Xr k=0

(−1)kdi1di2. . . dik ∂L

∂yσi1i2...ik are theEuler-Lagrange expressions. The (n+ 1)-form

Eλ=p1 is theEuler-Lagrange formassociated with λ.

By anautomorphismofY we mean a diffeomorphismα:W →Y, whereW ⊂Y is an open set, such that there exists a diffeomorphism α0 : π(W) X such that πα = α0π. If α0 exists, it is unique, and is called the π-projection of α. The r-jet prolongationofαis an automorphismJrα:Wr→JrY ofJrY, defined by

Jrα(Jxrγ) =Jαr0(x)(αγα−10 ).

Ifξ is aπ-projectable vector field onY, and αt is the local one-parameter group ofξwith projection α(0)t, we define ther-jet prolongationof ξto be the vector field JrξonJrY whose local one-parameter group isJαrt. Thus,

Jrξ(Jxrγ) =

½d dtJαr

(0)t(x)tγα−1(0)t)

¾

0

.

The chart expression forJrξcan be found in [8] or [9].

We now compute the Lie derivativeJrξλ. Choose to this purpose a Lepage equiv- alent ρof λ, and denote by s the order of ρ. Since λ = hρ, or, which is the same, Jrγλ=Jsγρfor all sections γ, we obtain

JrγJrξλ=JsγJsξρ=Jsγ(iJsξ+diJsξρ).

Omittingγ and using the Euler-Lagrange form we get

(4)

Jrξλ=hiJs+1ξEλ+hdiJsξρ.

This is thedifferential first variation formula; the first term on the right is theEuler- Lagrange term, and the second one is theboundary term.

An automorphismα:W →Y ofY is said to be aninvariant transformationof a formη∈spW, if

Jsαη=η.

We say that aπ-projectable vector fieldξis thegeneratorof invariant transformations ofη, if

Jsξη= 0.

The following simple consequence of the first variation formula is known as the Noether’s theorem. Let λ rn,XW be a Lagrangian, let ρ snW be a Lepage equivalent of λ, and let γ be an extremal. Then for any generator ξ of invariant transformations ofλ,

dJsγiJsξρ= 0.

An (n1)-formiJsξρis called theNoether’s currentassociated with a Lepage form ρand a vector fieldξ.

3 Frame bundle and its second jet prolongation

Let X be an n-dimensional smooth manifold, and let µ : F X X be the frame bundleoverX.F X has the structure of a right principalGln(R)-bundle. Recall that for every chart (U, ϕ),ϕ= (xi), onX, theassociated chartonF X, (V, ψ),ψ= (xi, xij), is defined byV =µ−1(U), and

xi(Ξ) =xi(µ(Ξ)), Ξj =xij(Ξ) µ

∂xi

x

,

where Ξ∈V,x=µ(Ξ), and Ξ = (x,Ξj). We denote by ykj theinverse matrixofxij. The right action F X ×Gln(R) 3 (Ξ, A) RA(Ξ) = Ξ·A F X is given by the equations

¯

xi=xi◦RA=xi, x¯ij=xij◦RA=xikakj, whereA=aij is an element of the groupGln(R).

For the formulation of variational principles on the frame bundles in this paper we need the r-jet prolongations of F X, the manifolds JrF X, where r = 1,2,3,4.

These manifolds are constructed from sections of the frame bundleF X in a standard way. We introduce basic concepts for J2F X, more general description of JrF X is available in [1]. To the charts (U, ϕ), and (V, ψ), introduced above, we associate a chart (V2, ψ2),ψ2= (xi, xij, xij,k, xij,kl), as follows. We denote byV2 the set of 2-jets of smooth frame fieldsU 3x→γ(x)∈V ⊂F X. IfJx2γ∈V2, we set

xi(Jx2γ) =xi(x), xij(Jx2γ) =xij(γ(x)),

xij,k(Jx2γ) =Dk(xijγϕ−1)(ϕ(x)), xij,kl(Jx2γ) =DkDl(xijγϕ−1)(ϕ(x)).

(5)

As usual, the 2-jets are equivalence classes of frame fields, which have the contact up to the second order, and have the canonicaljet prolongation x →J2γ(x) =Jx2γ of any frame fieldγ. The general linear group acts onJ2F X on the right by the formula Jx2γ·A=Jx2·A); the action is expressed by the equations

(3.1) x¯i=xi, x¯ij=xikakj, x¯ij,k=xim,kamj , x¯ij,kl=xim,klamj .

It is easy to determine the orbits of the action (Jx2γ, A)→Jx2·A). Denoting Γikp=−ypmxim,k, Γiklp=−ypmxim,kl,

we obtainGln(R)-invariant functions onJ2F X, andequations of Gln(R)-orbits Γikp=cikp, Γiklp=ciklp,

wherecikp, ciklp∈R are arbitrary numbers. The functions Γiklp are symmetric ink, l.

We have the following result.

Lemma 1.Every Gln(R)-invariant function onJ2F X depends onxi,Γikp,Γiklp. In other words Lemma 1 says that Gln(R)-invariant functions coincide with the functions on thebundle of second order connections C2X =J2F X/Gln(R) over X.

From equations (3.1) we can obtain an extension of Lemma 1 to differential forms.

Lemma 2. A k-form η on J2F X is Gln(R)-invariant if and only if it has an expression

η= ∆0+yqr11dxpq11rp11+yqr11yqr22dxpq11∧dxpq22rp11rp22 +. . .+yrq11yqr22. . . yrqkkdxpq11∧dxpq22. . .∧dxpqkkrp11rp22...r...pkk, where0, ∆rp11, ∆rp11rp22, . . . , ∆rp11rp22...r...pkk are arbitrary forms defined onC2X.

Let

ξ0=ξji Ã

∂aij

!

e

be a vector belonging to the Lie algebragln(R). Then the corresponding fundamental vector field onJ2F X is given by

(3.2) J2ξ=ξis Ã

xti

∂xts +xti,k

∂xts,k +xti,kl

∂xts,kl

! .

4 Reduction of the Euler-Lagrange equations

Using the results of Section 3, we determine in this section Gln(R)-invariant La- grangians on J1F X and J2F X. Then we give explicit expressions of the Euler- Lagrange forms, and the Noether’s currents associated with the Lepage equivalents Θλof these Lagrangians. Then we discuss consequences ofGln(R)-invariance of these

(6)

Lagrangians for the Euler-Lagrange equations. Our main tool is the first variation formula (Section 2).

Let us denote by Ψλ,ξ the Noether’s current associated with the Lepage form Θλ

(2.2), (2.3) and a vector fieldξ, and byωji the contact forms defined by ωij=dxij−xij,mdxm=dxij+xpjΓimpdxm.

Lemma 3.Let λ∈1n,XF X be a Lagrangian expressed byλ=0. (a)λisGln(R)-invariant if and only if Ldepends on xi,Γikj only.

(b)The Euler-Lagrange form of aGln(R)-invariant Lagrangian has an expression Eλ=yjl

Ã

−Γpqi ∂L

∂Γpql+ Γlpq ∂L

∂Γipq + 2L

∂xp∂Γipl

+(Γkmpr+ ΓkmqΓqpr) 2L

∂Γkmr∂Γipl

!

ωij∧ω0.

(c) If λ is Gln(R)-invariant, then the Noether’s current associated with the Poincar´e-Cartan form ofλand any fundamental vector field ξis given by

(4.1) Ψλ,ξ =−ξjmyjlxim ∂L

∂Γiklωk.

LetX be ann-dimensional manifold, letF X be the bundle of frames overX, and letµbe the bundle projection. Suppose that we have a Lagrangianλ∈1n,XF X and aµ-verticalvector fieldξ onF X. Then in our standard notation

(4.2) J1ξλ=iJ2ξEλ+hdiJ1ξΘλ, where Θλ is the Poincar´e-Cartan equivalent ofλ.

Theorem 1. Let λ∈1n,XF X be aGln(R)-invariant Lagrangian, letn≥2, and letγbe a section ofF X. The following conditions are equivalent.

(a)γ satisfies the Euler-Lagrange equations, Eλ◦J2γ= 0.

(b)For any chart(U, ϕ),ϕ= (xi), on X, and all j,k, there exist (n2)-forms ηjk such that

J1γ µ

yljxik ∂L

∂Γimlωm−dηkj

= 0.

Proof. By hypothesis, for any fundamental vector field ξ on F X, J1ξλ = 0.

Consequently, sinceξis alwaysµ-vertical, the first variation formula (4.2) reduces to

(4.3) iJ2ξEλ+hdΨλ,ξ= 0.

(7)

We can write this identity in a chart (U, ϕ), ϕ = (xi), on X. Using (3.2) we have, according to Lemma 3,

iJ2ξEλ=Eji(L)ξjkxikω0, where

Eji(L) = Ã

−Γpqi ∂L

∂Γpql + Γlpq ∂L

∂Γipq+ 2L

∂xp∂Γipl+ (Γkmps+ ΓkmqΓqps) 2L

∂Γkms∂Γipl

! ylj,

and the Noether’s current Ψλ,ξ is given by (4.1). It is convenient to denote ψmj =yjlxim ∂L

∂Γiklωk.

Then Ψλ,ξ=−ξjmψmj, and the first variation formula (4.3) can equivalently be written in the formEij(L)ξjkxikω0−ξjkhdψkj = 0. But the numbersξkj ∈Rare arbitrary, so we have

(4.4) Eij(L)xikω0−hdψjk= 0.

Suppose now that a sectionγsatisfies the Euler-Lagrange equations. Then the form Eji(L)xikω0 vanishes alongJ2γ, so we haveJ2γkj =dJ2γψjk = 0. Integrating we can find an (n2)-formηjk onU such that

(4.5) J1γψjk=kj.

Conversely, if a sectionγ satisfies condition (4.5), then by (4.4), γ is necessarily an extremal. 2

For second order Lagrangians onF X we have the following results.

Lemma 4.Let λ∈2n,XF X be a Lagrangian expressed byλ=0. (a)λisGln(R)-invariant if and only if Ldepends on xi,Γikj,Γiklj only.

(b)The Euler-Lagrange form of aGln(R)-invariant Lagrangian has an expression Eλ=yjl

Ã

−Γpqi ∂L

∂Γpql Γpqmi ∂L

∂Γpqml + Γlpq ∂L

∂Γipq +dp ∂L

∂Γipl Γlpqm ∂L

∂Γipqm

−2Γlpt µ

Γtqm ∂L

∂Γipqm+dq ∂L

∂Γipqt

−dpdq ∂L

∂Γipql

!

dxij∧ω0.

(c)IfλisGln(R)-invariant, then the Noether’s current associated with the Lepage form(2.3)and any fundamental vector fieldξ is given by

Ψλ,ξ =ξjmyljxim Ã

∂L

∂Γikl + Γqpi ∂L

∂Γqpkl + Γlpq ∂L

∂Γipkq +dp ∂L

∂Γipkl

! ωk.

Theorem 2. Let λ∈2n,XF X be aGln(R)-invariant Lagrangian, letn≥2, and letγbe a section ofF X. The following conditions are equivalent.

(8)

(a)γ satisfies the Euler-Lagrange equations, Eλ◦J4γ= 0.

(b)For any chart(U, ϕ),ϕ= (xi), on X, and all j,k, there exist (n2)-forms ηjk such that

J3γ Ã

yjlxik Ã

∂L

∂Γiml Γqpi ∂L

∂Γqpml Γlpq ∂L

∂Γipmq −dp ∂L

∂Γipml

!

ωm−dηkj

!

= 0.

Proof.The first variation formula for a second order Lagrangian λhas the form (4.6) J2ξλ=iJ4ξEλ+hdiJ3ξΘλ,

where Θλ is the Lepage equivalent ofλgiven by (2.3). Again, left hand side vanishes and formula (4.6) reduces to

iJ4ξEλ+hdΨλ,ξ= 0,

where the formsEλand Ψλ,ξ are given by Lemma 4. The rest of the proof is analogous to the Proof of Theorem 1. 2

Acknowledgments.The author is grateful to the Ministry of Education of the Slovak Republic (Grant VEGA No. 1/3009/06) and to the Czech Grant Agency (Grant No.

201/06/0922).

References

[1] J. Brajerˇc´ık, Higher order invariant variational principles on frame bundles, Ph.D. Thesis, Masaryk University, Brno, 2005.

[2] J. Brajerˇc´ık and D. Krupka,Variational principles for locally variational forms, J. Math. Phys. 46, 052903 (2005), 1-15.

[3] M. Castrill´on, P.L. Garc´ıa and T.S. Ratiu,Euler-Poincar´e reduction on principal bundles, Lett. Math. Phys. 58 (2001), 167-180.

[4] M. Castrill´on, T.S. Ratiu and S. Shkoller,Reduction in principal fiber bundles:

covariant Euler-Poincar´e equations, Proc. Amer. Math. Soc. 128 (2000), 2155- 2164.

[5] P.L. Garc´ıa,The Poincar´e-Cartan invariant in the calculus of variations, Sym- posia Mathematica 14 (1974), 219-246.

[6] I. Kol´aˇr, P.W. Michor and J. Slov´ak,Natural Operations in Differential Geome- try, Springer Verlag, Berlin, 1993.

[7] D. Krupka, A geometric theory of ordinary first order variational problems in fibered manifolds II. Invariance, J. Math. Anal. Appl. 49 (1975), 469-476.

[8] D. Krupka, Natural Lagrange structures, Differential Geometry, Banach Center Publications, Polish Scientific Publishers, Warsaw 12 (1984), 185-210.

[9] D. Krupka, Some Geometric Aspects of the Calculus of Variations in Fibered Manifolds, Folia Fac. Sci. Nat. UJEP Brunensis, University J.E. Purkynˇe, Brno, 14, 1973.

(9)

[10] D. Krupka,The geometry of Lagrange structures, Preprint series in Global Anal- ysis GA 7/1997, Silesian University, Mathematical Institute, Opava, 1997.

[11] D. Krupka and J. Janyˇska, Lectures on Differential Invariants, Folia Fac. Sci.

Nat. UJEP Brunensis, University J.E. Purkynˇe, Brno, 1990.

[12] D. Krupka and M. Krupka, Jets and contact elements, Proc. Semin. on Diff.

Geom., Mathematical Publications Vol. 2, Silesian University, Mathematical In- stitute, Opava (2000), 39-85.

[13] J. Mu˜noz Masqu´e and M. Eugenia Rosado Mar´ıa,Invariant variational problems on linear frame bundles, J. Phys. A: Math. Gen. 25 (2002), 2013-2036.

[14] D.J. Saunders,The Geometry of Jet Bundles, London Mathematical Society Lec- ture Note Series 142, Cambridge University Press, New York, 1989.

[15] A. Trautman,Invariance of Lagrangian systems, in: General Relativity, Papers in honor of J. L. Synge, Clarendon Press, Oxford (1972), 85-99.

Author’s address:

J´an Brajerˇc´ık

Department of Mathematics, University of Preˇsov, Ul. 17. novembra 1, 081 16 Preˇsov, Slovakia.

E-mail: [email protected]

参照

関連したドキュメント

Recently, a new FETI approach for two-dimensional problems was introduced in [16, 17, 33], where the continuity of the finite element functions at the cross points is retained in

The torsion free generalized connection is determined and its coefficients are obtained under condition that the metric structure is parallel or recurrent.. The Einstein-Yang

When P is an SI property, a much more efficient algorithm can be obtained by adjoining terms to both sides of the sequences, not just one side as in A 0... Then T 1 (P) is as

Keywords: Homotopy analysis method, Laplace transform, nonlinear Sys- tem of equations, Boundary value problems, Numerical methods..

It is suggested by our method that most of the quadratic algebras for all St¨ ackel equivalence classes of 3D second order quantum superintegrable systems on conformally flat

This paper targets engineers and practitioners who want a simple procedure to compute the mean performance measures of the Batch Markovian Arrival process (BMAP/G/1) queueing

Average Fraction Inspected (AFI) gives the expected fraction of inspected items to the number of products in the lot as a function of incoming quality, and, thus, is the most

In this paper, we discussed the second-order field equation associated to the Robertson-Walker space-time, that is, the generalized Klein- Gordon wave equation, using an