sign in sign up
<<
Home , Geometric calculus

arXiv:1910.11688v4 [math-ph] 6 Jan 2021 osre uniis nain agMlsconnections Yang–Mills invariant quantities: conserved 00MSC 2010 theory. Yang–Mills equation, Jacobi words Key ymtytasomtoso xrml n higher and extremals of transformations Symmetry ∗ orsodn Author Corresponding i .Abro1,113Trn,Iay e–mail: Italy, Torino, 10123 10, Alberto C. via aitosb rvn htapi ie yasmer fthe of symmetry a by of given symmetries pair study a and that proving fields by Jacobi variations to symmetr extremals relate We of mations laws. conservation shell’ ‘on generating rnfrain scntutdad sacs fsuy t e its study, space-t Minkowski of tained. on case connections a Yang–Mills invariant as s for two and, with constructed associated is current transformations conserved The current. served to faLgaga n yaJcb edo the of field Jacobi a by and Lagrangian a of ation h aeLgaga (with Lagrangian same the ecaatrz ymtytasomtoso arninex Lagrangian of transformations symmetry characterize We 54C teh,TeNtelns e–mail: Netherlands, The Utrecht, CD 3584 81T13,53Z05,58A20,58E15,58Z05. : ymtytasomto,cnevto a,scn variation, second law, conservation transformation, symmetry : eateto ahmtc,Uiest fUtrecht of University , of Department eateto ahmtc,Uiest fTorino of University Mathematics, of Department uaAccornero Luca aclaPalese Marcella l < s Abstract sascae iha o hl’con- shell’ ‘off an with associated is ) 1 [email protected] ∗ [email protected] s t aito of variation -th msi ob- is imes transfor- y xpression ymmetry l t vari- -th tremals higher L. Accornero and M. Palese 2

1 Introduction

The description of fundamental interactions in Physics as fields associated with the action of Lie groups on maps between has been the corner- stone of the last Century. Indeed within this picture fields are (local) maps between manifolds adapted to a fibration (which distinguishes independent from dependent variables and their peculiarity when changing coordinates), i.e. are sections of fibrations, having the additional structure of a bundle (fields take values in a which is the type fiber of the bundle). In particular, it is well known that, due to their invariance properties, physical fields can be described as sections of bundles associated with principal bun- dles, the configuration bundles are then the so-called gauge-natural bundle, see e.g. [8, 17]. It is noteworthy that the variational derivation of field equations is an intrinsic operation strictly related just to the fibration structure and its pro- longation up to a given order [8, 15, 33, 35]. This approach had several important developments, in particular when combined with invariance prop- erties (geometric formulation of the Noether Theorems, specifically). Furthermore important is now the possibility of a systematic formulation of higher variations (see Theorem 3.7 and Theorem 4.11 below), which can be interpreted as variations of suitable ‘deformed’ Lagrangians [2]. Combined with symmetry considerations this approach extends to field theory the somehow analogous concept of so-called Sarlet–Cantrijn higher- order (dynamical) Noether symmetries in Mechanics [32] (which we can think as a kind of higher order Noether–Bessel-Hagen symmetries of the Lagrangian [7, 11]). The present paper refers directly to canonical Noether symmetries, i.e. symmetries of the Lagrangian, which we can describe, roughly speaking, as dynamical Noether symmetries up to . Restricting to Mechan- ics, we recognize second order Sarlet–Cantrijn dynamical Noether symmetries up to divergences as related with second order canonical Noether currents in our approach. Higher variations are of interest in theoretical physics, in particular con- cerning variations of currents [13, 25]; for applications of the second variation in gravitational theory in this context see e.g. [14]. Lagrangian symmetries and symmetries of Euler–Lagrange equations have been called by Andrzej Trautman [35] invariant transformations and gener- alized invariant transformations, respectively, and they were characterized as particular kinds of what he called symmetry transformations, i.e. transfor- L. Accornero and M. Palese 3 mations of extremals into extremals of the same Euler–Lagrange equations. Indeed, it is well known that a symmetry of Euler–Lagrange equations (generalized invariant transformations) is also a symmetry transformation of their solutions (extremals), i.e. a transformation preserving the property of a field (a section of the configuration bundle on space-time) being an extremal [21, 35]. The inverse in general is not true: symmetry transformations of solutions of equations could not be symmetries of the equations themselves. A related result stating that a Lagrangian ‘dragged’ along symmetry transformations of its own extremals, has the same extremals as the original Lagrangian (and an inverse statement stating that a transformation dragging a given Lagrangian in a Lagrangian having the same extremals is a symmetry transformation of the extremals) was obtained in [23] (see Theorem 4.3 below). We focus on conservation laws by explicating the relation among higher variations of Lagrangians, symmetry transformations of extremals, Jacobi fields, and conserved currents. We characterize symmetry transformations of extremals as particular transformations of the Euler–Lagrange forms to source forms vanishing along extremals of the original Lagrangian and specif- ically as Jacobi fields along extremals (Theorem 4.5). Compared with generalized symmetry transformations (i.e. transforma- tions leaving invariant the Euler–Lagrange form of a Lagrangian) such trans- formations provide a weaker invariance property, since the Euler–Lagrange form is not invariant under their action, although it is transformed to a source form having the same extremals. Therefore the equations change, but the solutions of the one equation are also solutions of the second and vice versa. By explicating the relation among the variation of an Euler-Lagrange form with the second variation of a Lagrangian and with the Jacobi morphisms (Proposition 3.4 and Remark 4.14), we prove that with this sort of weaker invariance is anyway associated a conserved current, and in particular that this current can be identified as a Noether current for a Lagrangian ‘deformed’ by a symmetry transformation of extremals and associated with (or generated by) a symmetry transformation of extremals. More specifically, symmetry transformations of extremals generate conserved currents along the extremals themselves. Indeed Theorem 4.15 states the existence of a weak (i.e. along extremals) conservation law for any couple of (infinitesimal) generators of (vertical) symmetry transformations. As an explicit example, we write the expressions of the on shell conserved current generated by couples of symmetry transformations of extremals. L. Accornero and M. Palese 4

2 Contact structure, geometric and the ‘representation’ problem

Throughout this paper, we work with a fibration π : Y X, where Y is an m + n dimensional manifold and X is an n-dimensional→ manifold. When choosing coordinates, we will always pick fibered coordinates (xi,yσ) defined over open subsets π−1(U) Y , where U is an open subset of the base. We also set ds to be the local expression⊂ of a volume element dx1 ... dxn on X ∂ ∧ ∧ and dsi = ∂xi ds. Given this geometric setting, physical fields are encoded as local sections⌋ of the surjective submersion π : Y X. k → k Let Ωq(J Y ) denote the module of q-forms on the set J Y of equivalence classes of (local) sections of the fibration having a contact of order k in a point. Note that J kY as the structure of a differentiable manifold and the structure of a fibration π : J kY X, the prolongation of order k of k → π : Y X. A prolongation→ map jk assigns to (local) sections of the fibration π : Y k k → X (local) sections of the fibration πk : J Y X. If γ is a section of π, j γ is → k defined as the map assigning to x X the k-jet jx γ of σ at x. A differential q-form α Ω (J kY ) is called contact∈ if jkγ∗(α) = 0 for all sections γ of π. ∈ q It is easy to see that forms ω locally given as

ωσ = dyσ yσ dxi j1...jh j1...jh − j1...jhi for 0 h

π∗ (ρ)= p ρ + p ρ + + p ρ k+1,k 0 1 ··· q where p0ρ is a form that is horizontal on X (and so is often denoted by hρ) while piρ is an i-contact q-form, that is a form generated by wedge products L. Accornero and M. Palese 5 containing exactly i contact 1-forms. We remark that if q > n every q-form ρ is contact; then we call it strongly contact if pq−n = 0. The contact structure ∗ induces also the splitting of the exterior differential πk+1,kdρ = dHρ + dV ρ in q the so called horizontal and vertical differentials, given by dH ρ = P pldplρ l=0 q and dV ρ = P pl+1dplρ, respectively. l=0 According to [20] we define the formal with respect to the i- th coordinate, i = 1,...,n, by an abuse of notation also denoted by di, as an operator acting on forms. Explicitly, we require di to be the usual on 0-forms, to commute with the and to satisfy the Leibniz rule with respect to the wedge product. We see that q i dH ρ =( 1) diρ dx if ρ is a q-form. On the basis 1-forms associated to the − ∧ j σ σ σ σ contact structure we have didx = 0, diωj1...jr = ωj1...jri, didy = dyi . Notice k that di induces a vector field on J Y along πk+1,k; we still denote it by di and refer to it as the formal derivative. In the following a multi-index will be an ordered s-uple I =(i1,...is); the length of I is given by the number s; and an expression such as Ij denotes the multi-index given by the (s + 1)-uple (i1,...is, j). As much as the integration by parts procedure is concerned, we will use σ the local formula ωIi ds = dωI dsi. We also recall the properties σ σ ∂ ∧σ σ J− ∧ dJ ω = ωJ and ∂yν ωI = δν δI (where the Kronecker symbol with multi- J ⌋ indices has the obvious meaning: it is 1 if the multi-indices coincide up to a rearrangement and 0 otherwise). Finally, if ψ is a projectable vector field on Y (i.e. an infinitesimal au- tomorphism preserving the fibration), jkψ is the projectable vector field, defined on J kY , associated with the prolongation of the flow of ψ (see, e.g. [21, 33, 35]). The contact structure of jet prolongations enables to define an algebro- geometric object deeply related to the of variations: a differential sequence of sheaves made of equivalence classes of differential forms taking a variational meaning. We refer to [22], where the construction of a sequence of ‘variational sheaves’ can be found and to [20, 26] for the representation of finite order variational sequences. The concept of a sheaf is due to Leray; a classical reference on this topic is e.g. [6]. k k Let Ωq denote the sheaf of differential q-forms on J Y . It can be seen as a sheaf on Y ; in fact we assign to an open set W Y a form defined on ⊆ L. Accornero and M. Palese 6

−1 k k πr,0 (W ). We set Ω0,c = 0 and denote by Ωq,c the sheaf of contact q forms, for q n, or the sheaf{ of} strongly contact q-forms if q > n. The quotient ≤ sequence of the de Rham sequence of forms

0 R Ωk Ωk /Θk Ωk /Θk Ωk 0 , { }→ Y → 0 →···→ n n → n+1 n+1 →···→ N →{ } k k k k R where Θq = Ωq,c + dΩq−1,c, N = dim(J Y ) and Y is the constant sheaf over Y , is called the Krupka’s variational sequence of order k [22]. Let us k denote the quotient sheaves by q . Morphisms in this sequence are denoted k k V by q : q q+1 and they are quotients of the exterior differential i.e. ([Eρ]) =V [dρ→]. VBy this construction classes of forms modulo contact forms Eq are interpreted as differential forms relevant for calculus of variation (La- grangians, currents, source forms and so on); the Euler–Lagrange mapping can be identified with a morphism in the variational sequence. The represen- tation of the second variational derivative has been studied from this point of view [10, 12, 28, 29]. The interest of this construction in Physics is that the cohomology of the complex of global section of the variational sequence is the de Rham cohomology of Y [22]. Dealing with exact sequence of sheaves and resolu- tions enables to study cohomology obstructions to variational exactness of variationally closed forms and this turns out to be of interest in many dif- ferent areas of Physics; for example an obstruction to the existence of global extremals is related to the obstruction to the existence of global Noether– Bessel-Hagen currents [30]. Strictly related to concrete applications is then the so called representa- tion problem, which, roughly speaking, consists in showing that classes of forms, i.e. elements of the quotient groups r, can be associated with global Vq differential forms. By the intrinsic geometric structure of the on finite order prolongations of fibrations, indeed, it is possible to define an operator (called representation mapping) which takes a differential forms on the prolongation of order r and associate to it a differential form on a certain prolongation order s r, having a meaning in the Lagrangian r≥ r s s formalism for field theory, i.e. Rq : Ωq Ψq, with Ψq an abelian group r → r of forms of order s, such that ker Rq = Θq. It provides an isomorphism r s r r q = Ψq = Rq(Ωq). V ∼ r For q n, Rq can be taken to be simply the ‘horizontalization’ h = p0. For q n≤+ 1, it is the image of an operator denoted by , which will be ≥ I suitably defined below and which reflects in an intrinsic way the procedure L. Accornero and M. Palese 7

k k of getting a distinguished representative of a class [ρ] Ωq /Θq for q > n by ∈ k applying to ρ the operator pq−n and then factorizing by Θq , see e.g. [26]. In this paper we will refer to the interior Euler operator defined within the finite order variational sequence according to [19, 20] and applied to the representation of variational Lie according to [26]. Definition 2.1 In the following, differential forms which are ωσ generated l-contact (n + l)-forms will be called source forms. Now define locally the map : Ωr Ω2r+1 by I n+k → n+k r 1 σ 1 σ |I| ∂ (ρ)= ω Iσ = ω X( 1) dI ( pkρ) . I k ∧ k ∧ − ∂yσ ⌋ |I|=0 I For a given ρ, (ρ) is a source form of degree n + k and it is by construction a k-contact form.I It turns out that, if ρ is global, (ρ) is a globally defined form; for a proof, see [20]. I In view of a characterization of Noether currents, we study the difference between (π )∗(p ρ) and (ρ). In particular, we define the residual op- 2r+1,r+1 k I erator by the following decomposition formula which is in fact a geometric integrationR parts formula (π )∗(p ρ)= (ρ)+ p dp (ρ) . (1) 2r+1,r+1 k I k kR Note that, although the decomposition above has a global meaning, (ρ) is a strongly contact (n + k 1)-form defined only locally. R − Example 2.2 Following [20] we can characterize (ρ) in local coordinates. For k 1, if ΨI are (k 1)-contact (n+k 1)-formsR and if ωσ are local gener- ≥ σ − − I ators of contact 1-forms, up to pull-backs, we can write (a sort of integration by parts on formal derivatives of forms) r r σ I σ I pkρ = X ωI Ψσ = X dI (ω ζσ)= (ρ)+ pkdpk (ρ) , ∧ ∧ I R |I|=0 |I|=0

r−|I| I |J| |I|+|J| JI with ζσ = P ( 1) |J| dJ Ψσ . The first term gives us the Euler–Lagrange |J|=0 − σ I I I form, while by rewriting ω ζσ = Φ ds, for suitable k-contact k-forms Φ on J 2rY , we get ∧ ∧ r r−1 σ I k Ij X dI(ω ζσ)= dH (X( 1) dI Φ dsj)= dH (ρ) . ∧ − ∧ R |I|=1 |I|=0 L. Accornero and M. Palese 8

This local expressions for (ρ) will be exploited in Example 5.2 for the case k = 1, specifically for concreteR 1-contact (n + 1)-forms ωσ ΨI associated I ∧ σ with the exterior differential of a suitably ‘deformed’ Yang–Mills Lagrangian. We will write explicitly the forms ΦIj relative to this Lagrangian. Combined with results of Theorem 4.15, this approach will enable us to obtain explicit conserved currents associated with symmetry transformations of Yang–Mills extremals on Minkowski space-times.

3 Higher variations and related currents

The representation by the horizontalization h and by the interior Euler op- erator (also called the Takens representation [26]) defines a sequence of I sheaves of differential forms (rather than of classes of differential forms), such that both the objects and the morphisms have a straightforward inter- pretation in the calculus of variations. We can obtain formulae for (higher) variations of a Lagrangian based on an iteration of the first variation formula expressed through this representation.

3.1 Noether currents The formulation of the First Noether Theorem [24] is concerned with the rep- resentation of variational Lie derivatives of classes of degree n, which illus- trates the relation between the interior Euler operator, the Euler–Lagrange operator and the exterior differential, as well as the emerging of the di- vergence of the Noether currents by contact decompositions and geometric integration by parts formulae. In the following, for any n-form ρ, (dρ)= (dhρ)= d(hρ) is the Euler– I I I Lagrange form En(hρ), while hdhµ = p0dp0µ is the horizontal differential dH (hµ), which can be recognized as a (for the notation and the interpretation in the context of geometric calculus of variations more details can be found e.g. in [26]).

Theorem 3.1 For any n-form ρ and for any π-projectable vector field ψ on Y , we have, up to pull-backs by projections,

r+1 L r+1 hρ = ψ d(hρ)+ d (J ψ p + ψ hρ) (2) J ψ V ⌋I H V ⌋ dV hρ H ⌋ where p = p (dhρ). dV hρ − 1R L. Accornero and M. Palese 9

The formula above has been first obtained by Noether in the proof of her celebrated First Theorem (see the original Noether paper in the historical survey [18]). This suggest the definition of a Noether current. Definition 3.2 The Noether current for a Lagrangian λ associated with ψ is defined as ǫ (λ)= J r+1ψ p + ψ λ . ψ V ⌋ dV λ H ⌋ The term p = p (dλ) is called a local generalized momentum. dV λ − 1R It should be stressed that a Noether current is defined for any projectable vector field, independently from it being a Lagrangian symmetry or not. When it is not a symmetry of course the Noether current is not conserved along critical sections. A generalization of formula (2) to class of degree greater or lower than n has been obtained [7, 26, 27].

3.2 Higher Noether currents We now tackle a systematic formulation of higher variations, interpretable as variations of suitable ‘deformed’ Lagrangians [2]; combined with symmetry considerations this approach extends to field theory the concept of so-called higher-order Noether symmetries in Mechanics developed in [32]. Now we obtain a formula for the second variation, which will be further exploited in section 4. We note that LJr+1ψhρ = hLJrψρ, and then apply a standard inductive reasoning. Of course, the iterated variation is pulled-back up to a suitable order, in order to suitably split the Lie derivatives [2]. Theorem 3.3 For any n-form ρ and any pair of π-projectable vector fields ψ1 and ψ2, we have, up to pull-backs by projections,

L r+1 L r+1 hρ = ψ d(ψ d(hρ))+ (3) J ψ2 J ψ1 2,V ⌋I 1,V ⌋I +d ǫ (ψ d(hρ)) + d ǫ (d ǫ (hρ)) H ψ2 1,V ⌋I H ψ2 H ψ1 where we define the following (higher) Noether currents associated with ψ2 for the respective new Lagrangians:

r+1 ǫψ2 (ψ1,V d(hρ)) = ψ2,H ψ1,V d(hρ)+ J ψ2,V pdV ψ1,V ⌋Id(hρ) , (4) ⌋I ⌋ ⌋I r+1 ⌋ ǫψ2 (dH ǫψ1 (hρ)) = ψ2,H dH (J ψ1,V pdV hρ + ψ1,H hρ)+ (5) r+1⌋ ⌋ ⌋ +J ψ2,V pd d (Jr+1ψ ⌋p +ψ ⌋hρ) . ⌋ V H 1,V dV hρ 1,H L. Accornero and M. Palese 10

Note that the expression (3) is given in terms of and . Related to this formula is an identity which willI suggestR the definition of the Jacobi morphism, with a look to a specific characterization of symmetry transformations of extremals (see Definition 4.6). Let then ψ1, ψ2 be vertical vector fields. We note that, due to the ex- actness of the representation sequence and by linearity of the Lie derivative, (for s a suitable prolongation order) we can write

s J ψ L s d(hρ)= ψ d(ψ d(hρ))= (6) 1⌋ J ψ2 I 1⌋I 2⌋I = L s L s hρ [ψ , ψ ] d(hρ) d ǫ (d ǫ (hρ)) . J ψ2 J ψ1 − 2 1 ⌋I − H ψ2 H ψ1 From (3) we get then the following identity.

Proposition 3.4 For every pair of vertical vector fields ψ1 and ψ2 it holds

ψ d(ψ d(hρ)) ψ d(ψ d(hρ))= (7) 1⌋I 2⌋I − 2⌋I 1⌋I = [ψ , ψ ] d(hρ)+ d (ǫ (ψ d(hρ))) . 1 2 ⌋I H ψ2 1⌋I

Note that, being the vector fields vertical, here we have ǫψ2 (ψ1 d(hρ)) = r+1 ⌋I J ψ2 pdV ψ1⌋Id(hρ). Note also that this current is the Noether current for the ‘deformed’⌋ Lagrangian ψ d(hρ) and associated to ψ . 1⌋I 2 Remark 3.5 As we already mentioned in the Introduction there exists a concept of a higher order Noether symmetry referring actually to a higher order generalization of a ‘Noether symmetry’ intended as a symmetry of the exterior differential of the Poincar´e–Cartan equivalent of a given Lagrangian, i.e. Ψ is called a ‘Noether symmetry’ if LΨdθ = 0. In particular, we refer to the generalization due to Sarlet–Cantrijn [32], and in the following we shall call a ‘Noether symmetry’ according to them as a Sarlet–Cantrijn symmetry. We stress that this symmetry differs from a Noether symmetry as a sym- metry of the Lagrangian (due to the fact that the Poincar´e–Cartan equivalent of L differs for a 1-contact term, let us call it Ω), which is the original meaning used also by Emmy Noether who referred to symmetries of the action, i.e. of the Lagrangian. As well known symmetries of the Lagrangian are also sym- metries of the Euler–Lagrange form (which can be expressed, in terms of the differential of the Poincar´e–Cartan equivalent as as p1dθ), but the converse is not true in general. Indeed symmetries of the Euler–Lagrange form are generalized symmetries of the Lagrangian, i.e. symmetries up to a horizontal L. Accornero and M. Palese 11 differential. Therefore, Sarlet–Cantrijn symmetries, which are symmetries of the Poincar´e–Cartan equivalent up to a differential can be identified as a kind of generalized symmetries. Accordingly higher Noether symmetries and currents in this paper can be compared with Sarlet–Cantrijn ones.

Example 3.6 Let us now denote by θ = L + Ω the Poincar´e–Cartan equiv- alent of the Lagrangian L = hρ. Let us for a moment skip the prolongation symbols to simplify the notation; we have the following. Let Ψ denote a projectable vector field and let LΨLΨdθ = d(Ψ d(Ψ dθ)) = 0. On the one hand, by the naturality of the Lie derivative and by⌋ Equ⌋ ation (6) we have

LΨLΨdθ = dLΨLΨθ = = d[Ψ d(Ψ dL)+ d ǫ (Ψ dL)+ d ǫ (d ǫ (L)) + L L Ω] . V ⌋I V ⌋I H Ψ V ⌋I H Ψ H Ψ Ψ Ψ Let us take the quotient modulo the contact structure Θ, dL L θ. Since I Ψ Ψ dL L Ω= L L dΩ = 0, being Ω a contact form and being dΩ also in the I Ψ Ψ Ψ ΨI contact sheaf Θ. Thus we have dLΨLΨθ = dη, with η =ΨV d(ΨV dL)+ d ǫ (Ψ dL)+ d ǫ (d ǫ (LI)). I ⌋I ⌋I H Ψ V ⌋I H Ψ H Ψ Note that under the assumption that LΨLΨdθ = 0, η is a d-closed form, i.e. dη = 0, then by the exactness of the variational sequence,I we have I locally ΨV d(ΨV dL)= dH (G [ǫΨ(ΨV dL)+ ǫΨ(dH ǫΨ(L))]). On the⌋I other hand,⌋I following− [32], let β⌋I= Ψ dα, where α = Ψ dθ. We have L L θ = β + d(Ψ d(Ψ θ)) = β + dξ; and from⌋ the above we⌋ also get Ψ Ψ ⌋ ⌋ dη = d(β + dξ)= d(β + dV ξ). I WeI now elaborateI and compare these two issues. Indeed, from Proposition 3.4 we have the identity d ǫ (Ψ dL) = 0. Thus H ΨV V ⌋I Ψ d(Ψ dL)= d (G Ψ Ψ dL ǫ (d ǫ (L))]) . V ⌋I V ⌋I H − H ⌋ V ⌋I − Ψ H Ψ Furthermore, if we assume ΨV to be such that ΨV d(ΨV dL)=0(i.e. to be a Jacobi field, see later) then we get the conservation⌋I law⌋I d (G Ψ Ψ dL ǫ (d ǫ (L))) = 0 , H − H ⌋ V ⌋I − Ψ H Ψ which along an extremal reads d (G ǫ (d ǫ (L))) = 0 . H − Ψ H Ψ

On the other hand since LΨLΨdθ = d(Ψ d(Ψ dθ)) = 0 then dβ = 0, therefore locally β = dF [32]. ⌋ ⌋ L. Accornero and M. Palese 12

By taking the horizontal part we have hβ = dH F . But from the above dG = η = β + dξ thus we can take G = F + ξ, and we have locally that, for ΨV a Jacobi field along an extremal d F = d (ǫ (d ǫ (L)) ξ)= d (ǫ (d ǫ (L)) Ψ d(Ψ θ) . H H Ψ H Ψ − H Ψ H Ψ − ⌋ ⌋ Now, up to pull-backs and jet prolongations of the vector field Ψ, we have

dH F = dH (ΨH dH (ΨV pd L +ΨH L)+ΨV pd d (Ψ ⌋p +Ψ ⌋L) + ⌋ ⌋ V ⌋ ⌋ V H V dV L H Ψ (d (Ψ p +Ψ L)) Ψ (d (Ψ p +Ψ L)) + − H ⌋ H V ⌋ dV L H ⌋ − H ⌋ V V ⌋ dV L H ⌋ Ψ (d (Ψ p +Ψ L)) Ψ (d (Ψ p +Ψ L))) = − V ⌋ H V ⌋ dV L H ⌋ − V ⌋ V V ⌋ dV L H ⌋ dH (ΨV pd d (Ψ ⌋p +Ψ ⌋L) ΨH (dV (ΨV pd L +ΨH L)) + ⌋ V H V dV L H − ⌋ ⌋ V ⌋ Ψ (d (Ψ p +Ψ L)) Ψ (d (Ψ p +Ψ L))) = − V ⌋ H V ⌋ dV L H ⌋ − V ⌋ V V ⌋ dV L H ⌋ = dH (ΨV pd d (Ψ ⌋p +Ψ ⌋L) ΨV (dV (ΨV pd L +ΨH L)) ⌋ V H V dV L H − ⌋ ⌋ V ⌋ = d (ǫ (d ǫ (L)+Ψ p Ψ (d (Ψ p +Ψ L)) . H ΨV H ΨV V ⌋ dV dH (ΨH ⌋L) − V ⌋ V V ⌋ dV L H ⌋ which explicates the relationship between Sarlet–Cantrijn second-order Noether conserved current and the conserved current along extremals ǫΨV (dH ǫΨV (L)), for ΨV a Jacobi field; see later Equation (14) combined with Proposition 3.4. The last term can be shown to vanish under the horizontal differential [7, 26, 27], thus in the case of a vertical Sarlet–Cantrijn symmetry the con- served current F essentially coincides with the Noether conserved current

ǫΨV (LΨV (hρ)) up to a horizontal differential. By Theorems 3.1 and 3.3 formulae for the higher variations of hρ are obtained [1]. Theorem 3.7 Let ρ be an n-form on J kY . Consider the Lagrangian hρ and take l variation vector fields ψ1,...,ψl. Define recursively a sequence rl by

rl =2rl−1 +1, r0 = r We have ∗ r+1 r+1 (πrl,r+1) (LJ ψl ...LJ ψ1 hρ)= =ψ d(ψ d(...ψ d(ψ d(hρ)) ... )+ l,V ⌋I l−1,V ⌋I 2,V ⌋I 1,V ⌋I +dH ǫψ (ψl ,V d(...ψ ,V d(ψ ,V d(hρ)) ... )+ l −1 ⌋I 2 ⌋I 1 ⌋I (8) +d ǫ (d ǫ (ψ d(... (ψ d(hρ)) ... )+ H ψl H ψl−1 l−2,V ⌋I 1,V ⌋I ...

+dH ǫψl (dHǫψl−1 dH (...dH ǫψ2 (dH ǫψ1 (hρ)) ... ) . L. Accornero and M. Palese 13

Proof. The proof is a straightforward induction using as base step the case l =1 or l = 2. Taking into account the exactness of the representation sequence, the inductive step follows easily.

Remark 3.8 By means of a recursive application of (2) or (3), the currents which appear in formula (8) can be worked out more explicitly and charac- terized as Noether currents similarly to the expressions in (4) and (5).

The variation of any order of a Lagrangian hρ is a horizontal form, i.e. again a Lagrangian, therefore we can express its variation by means of for- mula (2). On the other hand formula (8) gives us the possibility to investigate how to relate the symmetries of a variation of hρ to hρ itself (see Theorem 4.11 below).

4 Symmetry transformations of extremals and conserved currents

Let hρ be a Lagrangian of order r +1 on Y . Definition 4.1 A (local) section γ is an extremal of hρ if it satisfies

d(hρ) J 2r+1γ =0 . I ◦ Let now φ be an automorphism of Y (i.e. a transformation preserving the r+1 fibration) with projection φ0, and let J φ be its prolongation.

Definition 4.2 The automorphism φ is a symmetry transformation of an extremal γ, if the section φ γ φ−1 is also an extremal. i.e. ◦ ◦ 0 d(hρ) J 2r+1(φ γ φ−1)=0 . I ◦ ◦ ◦ 0 A π-projectable vector field ψ is the generator of symmetry transforma- tions of γ, if its local one-parameter group of transformations is a flow of symmetry transformations of γ. It can be shown that a symmetry of d(hρ) is also a symmetry transformation of every extremal γ [35, 21]. I According with the above references, the following relates symmetry trans- formations of extremals with projectable vector fields dragging hρ in such a way that LJr+1ψhρ admits the same extremals. L. Accornero and M. Palese 14

Theorem 4.3 Let hρ be a Lagrangian of order r+1 and let γ be an extremal. Then a π-projectable vector field ψ generates symmetry transformations of γ if and only if

2r+1 d(L r+1 hρ) J γ =0 , I J ψ ◦ Remark 4.4 We make now an observation which will have a fundamental consequence when related to Theorem 3.3. Indeed, we note that, being the Lie derivative a natural operator, it holds LJ2r+1ψ d(hρ)= d(LJr+1ψhρ) and ψ generates symmetry transformations of γ if andI only if I

2r+1 (L 2r+1 d(hρ)) J γ =0 . J ψI ◦ We thus characterize vertical symmetry transformations of extremals as par- ticular transformations of the Euler–Lagrange forms to source forms vanish- ing along extremals of the original Lagrangian. We are now able to state our first main result, which is the premise for the next fundamental step: to characterize vertical symmetry transformations of extremals specifically as Jacobi fields along extremals (see Theorem 4.7 and Remark 4.14 below). We do this basically by expressing the Lie derivative of Euler–Lagrange forms in terms of the second variation (see also [28]).

Theorem 4.5 Let hρ be a Lagrangian of order r +1, let γ be an extremal. Then a vertical vector field ψ generates vertical symmetry transformations of γ if and only if

d(ψ d(hρ)) J 4r+1γ =0 . I ⌋I ◦ Proof. The result follows from Theorem 4.3 and Remark 4.4 by using the identity (6) which, we stress, holds true for any vertical vector field ψ1. We focus on higher order Noether currents and in particular on currents associated with the infinitesimal second variation formula (3) in a specific way. Roughly speaking, up to horizontal differentials, the second variation (generated by vertical vector fields) of a Lagrangian λ is the Jacobi mor- phism (see [15] for first order field theory; see also [12]). Here we define the Jacobi morphism within the representation sequence, i.e. by the interior Euler operator. L. Accornero and M. Palese 15

Definition 4.6 Let XV (Y ) be the space of vertical vector fields on Y . The map

: Ωr (J rY ) X∗ (J 2r+1Y ) X∗ (Y ) Ωr (J rY ) (9) J n,X → V ⊗ V ⊗ n,X λ : d( d(λ)) (10) 7→ • ⌋I • ⌋I is called the Jacobi morphism associated with the Lagrangian λ .

The Jacobi morphism is self-adjoint along critical sections of a Lagrangian field theory of any order (see also [2, 12]). This is a property of great impor- tance in physical applications.

Theorem 4.7 For any pair of vertical vector fields ψ1, ψ2 on Y , we have

J 2r+1ψ (J 2r+1ψ d (dλ))=0 . 2⌋I 1⌋ I Along extremals the Jacobi morphism is self adjoint.

Indeed we have

2r+1 |J| ρ ∂Eρ(λ) σ d(ψ d(λ)) = X ( 1) dJ (ψ )ω ds = (11) I ⌋I − ∂yσ ∧ |J|=0 J

2r+1 σ ∂Eρ(λ) ρ = X dJ ψ ω ds . (12) ∂yσ ∧ |J|=0 J

In the following we use the notation (λ) for short to denote d(ψ d(λ)). Jψ I ⌋I Definition 4.8 Let λ be a Lagrangian of order r. A Jacobi field for the Lagrangian λ is a vertical vector field ψ that belongs to the kernel of the Jacobi morphism, i.e. satisfying the Jacobi equation for the Lagrangian λ

(λ)=0 . Jψ

Remark 4.9 The Jacobi morphism ψ(λ), evaluated along an extremal γ, depends only on the values of the vectorJ field ψ along γ. A Jacobi equation along an extremal is then well defined; we call its solutions the Jacobi fields along an extremal γ. L. Accornero and M. Palese 16

Remark 4.10 Equation (12) provides the ‘adjoint expression’ for the Jacobi equation along extremals; it can be of use in order to obtain an easier char- acterization of the kernel of the Jacobi morphism in practical computations. Notably here it will be used in order to calculate the conserved current associated with invariant Yang–Mills connections (see Example 5.2 below). See also [31] for an explicit application in SU(3)-Yang–Mills theories in the context of a variationally featured symmetry breaking in view of a canoni- cal characterization of confinement phases in non-abelian gauge theories [34].

4.1 Jacobi fields and higher conservation laws We observe that the Jacobi equation for variations of hρ can be expressed in terms of hρ. In fact, just using the exactness of the representation sequence, we have

d(ψ d( L r+1 ...L r+1 hρ)) = d(ψ d(ψ d(...ψ d(hρ) ... ))) . I ⌋I ⌋ J ψs J ψ1 I ⌋I s⌋I 1⌋I The application of Theorem 4.15 to an iterated variation of a Lagrangian give results that are relevant for the Lagrangian itself; in fact, using (8) we can relate the Noether current of the s-th variation with Noether currents of lower order variations. More precisely, we can state the following important original result. Theorem 4.11 If we take a symmetry of an (l 1)-th variation of hρ and we suppose that the s-th variation (s

r+1 r+1 dH ǫψl (LJ ψl−1 ...LJ ψ1 hρ)=0 , with some terms that vanish separately. In fact, applying the definition of Jacobi field, we get ψ d(ψ d(...ψ d(ψ d(hρ)) ... )=0 , l⌋I l−1⌋I 2⌋I 1⌋I d ǫ (ψ d(...ψ d(ψ d(hρ)) ... )=0 , H ψl l−1⌋I 2⌋I 1⌋I ... d ǫ ...d ǫ (ψ d(... (ψ d(hρ)) ... )=0 . H ψl H ψs+2 s+1⌋I 1⌋I Then the statement follows easily using (8) (Theorem 3.7). L. Accornero and M. Palese 17

Remark 4.12 The previous result is a strong conservation law(i.e. it holds along any section, not necessarily an extremal) : the conserved current is the (l s 1)-th variation of the horizontal differential of the Noether current for− the−s-th variation of hρ. The result is not trivial because we are not assuming that ψs+1 is a symmetry of the s-th variation.

Remark 4.13 We can write formula (8) in terms of Jacobi morphisms:

∗ r+1 r+1 (πrl,r+1) (LJ ψl . . .LJ ψ1 hρ)= =ψ (ψ ... (hρ) ... )+ l,V ⌋Jψl−1,V l−2,V ⌋Jψl−3,V +d ǫ (ψ (ψ ... (hρ) ... ))+ H ψl l−1,V ⌋Jψl−2,V l−3,V ⌋ +d ǫ (d ǫ (ψ (... (hρ) ... )))+ H ψl H ψl−1 l−2,V ⌋Jψl−3,V + + d ǫ (d ǫ d (...d ǫ (hρ) ... )) . ··· H ψl H ψl−1 H H ψ1 Remark 4.14 Note that by Theorem 4.5, and by Theorem 4.7, Jacobi fields along extremals are vertical symmetry transformations of extremals and vice versa.

Our characterization of (vertical) symmetry transformations of extremals as Jacobi fields along extremals is motivated by the fact that there can ex- ist conservation laws associated with symmetry transformations, which in principle are different from the Noether or the Noether–Bessel-Hagen con- servation laws associated with symmetries or generalized symmetries of a Lagrangian λ, in particular we stress once more that all symmetries of equa- tions are also symmetry transformations of extremals, but the converse is not true in general. Indeed, we have the following our further main result.

Theorem 4.15 Let ρ be an n-form on J r−1Y and hρ the associated La- r grangian on J Y . Let ψ1 and ψ2 on Y be two generators of vertical sym- metry transformations of extremals. Then, along extremals of hρ, the weak conservation law holds true:

d ǫ (ψ d(hρ))=0 . (13) H ψ2 1⌋I Proof. Indeed, by Theorem 4.5, the two generators of symmetry trans- formations ψ and ψ are also Jacobi fields, i.e. they must satisfy (hρ) = 0, 1 2 Jψi for i =1, 2. Therefore from (7), since also [ψ2, ψ1] d(hρ) vanishes along ex- tremals, we get the result. ⌋I L. Accornero and M. Palese 18

For the interpretation of this current as a Noether current for a ‘deformed’ Lagrangian, see the note at the end of Proposition 3.4.

Remark 4.16 Suppose that ψ2 is a symmetry of the first variation of hρ generated by ψ and that ψ and ψ satisfy ψ (hρ) = 0, then we have a 1 1 2 2⌋Jψ1 strong conservation law:

dH ǫψ2 (LJrψ1 hρ)=0 . (14) Now, it is clear that along extremals, taking two vertical symmetry transfor- mations ψ1 and ψ2, we get two separated weak conservation laws; see also [2].

5 Conserved currents for invariant Yang–Mills connections

Let us consider a Yang–Mills theory [36] on the bundle (CP ,π,M) of principal connections with structure bundle (P,p,M,G), G being a semi-simple group. Lower Greek indices label space-time coordinates, while capital Latin indices label the Lie algebra g of G, then, on the bundle CP , we introduce coordinates µ A (x ,ωσ ). Let δ be the Cartan-Killing metric on the Lie algebra g, and choose a δ-orthonormal basis TA in g; the components of δ will be denoted δAB. The Yang-Mills Lagrangian is locally expressed by 1 λ = F A gµρgνσF B δ √gds , YM −4 µν ρσ AB where g stands for the absolute value of the determinant of the metric gµν , A A A A A B C cBC are the structure constants of g, Fµν = ων,µ ωµ,ν + cBC ωµ ων is the so A A − called field strength, and we set ωµ,ν = dνωµ . From now on, we assume the metric η to be Minkowskian; in this case the Euler–Lagrange expressions for the Yang–Mills Lagrangian are explicitly written as Eν = δ ηλµηǫν(ωA ωA + cA ωZ ωD + cA ωZωD )+ (15) B BA ǫ,λµ − λ,ǫµ ZD λ,µ ǫ ZD λ ǫ,µ +ηλµηǫνδ (ωD ωD + cD ωEωF )cA ωZ . DA ǫ,λ − λ,ǫ EF λ ǫ BZ µ Let (φa) be a set of coordinates on the group G. A vertical vector field over Z ∂ CP has the form ψ = ψ Z and its components satisfy the transformation σ ∂ωσ L. Accornero and M. Palese 19

′B B A µ B rule ψν = AdA(φ)ψµ J ν where AdA(φ) is the adjoint representation of G µ on g and J ν denotes the inverse of the matrix of the change of coordinates in the base space. Let L(M) be the frame bundle of M, V = g Rn. Let ⊗µ us denote by the covariant derivative corresponding to Ω = dx (∂µ + cB ψDωA∂σ )∇ the connection induced on the bundle (P L(M)) ⊗ V of AD σ µ B ×M ×λ vertical vector fields over CP , where the representation λ is comes from by the transformation rules above (see [16, 17, 9]). By some careful manipulations (see [2] for details), Theorem 4.7 and specifically formula (12) provide the Jacobi equation along extremals for this D kind of Yang-Mills theory. In particular, due to the antisymmetry of Fβσ in the lower indices, it splits in the antisymmetric and symmetric parts ην[σηβ]α ψA ψA δ + F D δ cA ψZ =0 , (16) ∇β ∇α σ −∇σ α  BA βσ AD BZ α ην(σηβ)α ψA ψA δ =0 . ∇β ∇α σ −∇σ α  BA for any pair (ν, B); hereafter the brackets () and [] in the superscripts denote symmetrization and anti-symmetrization, respectively. Note that the left hand side of these equations are the analogous, for a Minkowskian metric, of the classical expression for the Jacobi operator for Yang–Mills theories on different backgrounds, see e.g. [3, 5], and it reproduces results for first order non regular Lagrangians [15]. A A A In order to avoid notational confusion, let χµ , χµ,ν, χµ,νρ,... denote gen- erators of contact forms. Remark 5.1 According with our results, the solutions ψ of the above equa- tions are the generators of symmetry transformations of Yang–Mills ex- tremals ω, i.e. if ψ is a solution of the above equation, then the source ν B ν form LJ3ψ(EB χν ds), where EB are given by (15), also vanishes along the same extremals, i.e∧.

ν B 3 (L 3 (E χ ds)) J ω =0 . J ψ B ν ∧ ◦ Here, by a slight abuse of notation, we denoted by ω a section of the bundle (CP ,π,M) which is an extremal, i.e. a Yang–Mills connection. As we already mentioned, compared with transformations leaving invari- ν B ant the Euler–Lagrange form EB χν ds, such transformations are involved with a weaker invariance property, since∧ the Euler–Lagrange form is not in- variant under their action, but it is transformed to a source form having the same extremals. Note that indeed the Yang–Mills extremals ω are also solutions of the equation above and vice versa. L. Accornero and M. Palese 20

Example 5.2 As an illustration of the application of our main result, The- orem 4.15, we determine now the conserved current associated to such a weaker invariance property. We write down explicitly the current for two given generators of vertical symmetry transformations ψ and ψ˜, solutions of equation (16) (for details we refer to [1, 2], where computations are made for Jacobi fields along extremals). Being the vector fields vertical, from Theorem 4.15, equation (13), and according to Theorem 3.1, the conserved current along an extremal is defined ν through p1 (d(ψ d(λYM )). Recalling that EB are coordinate expression of the Euler–LagrangeR ⌋I form according to (15), we apply the coordinate charac- terization of the residual operator (given in Example 2.2) to the form

B ν ∂ψν ν B ∂EB Z d(ψ d(λYM ))=( Z EB + ψν Z )χρ ds + ⌋I ∂ωρ ∂ωρ ∧ ν ν B ∂EB Z B ∂EB Z +(ψν ∂ωZ )χρ,ξ ds +(ψν ∂ωZ )χρ,ξτ ds . ρ,ξ ∧ ρ,ξτ ∧ 2 A µ,I By suitably rewriting the above in the form P dI(χµ ζA ), we can easily |I|=0 ∧ obtain the local expression for (d(ψ d(λYM ))), thus obtaining the con- served current we are looking for:R ⌋I

ρ[ξ σ]ν A D B Z Z B ǫ ˜(ψ d(λ )) = [η η δ c ω (ψ ψ˜ ψ ψ˜ )+ ψ ⌋I YM BA ZD σ ν ρ − ρ ν (ηξσηρν ηρ(σηξ)ν)(ψZ (ψ˜Bδ ) ψ˜Z (ψBδ ))]ds . − ν ∇σ ρ BZ − ρ ∇σ ν BZ ξ Remark 5.3 It is noteworthy that, by Proposition 3.4 and in particular by Remark 4.14, here the existence and the meaning of the above current (also appeared in [2] in relation with Jacobi fields) is understood under a new light, definitely relevant from a physical point of view. In the present paper we clarify that such a conservation law emerges by an invariance property of the set of extremals and, moreover, that the associ- ated conserved current can be interpreted as a very specific kind of Noether current, the existence of which is related with a wide class of symmetry transformations. Indeed, we proved that this current can be identified as the Noether current for the Yang–Mills Lagrangian ‘deformed’ by the symme- try transformation of extremals ψ and associated with (or generated by) the symmetry transformation of extremals ψ˜.

Remark 5.4 We note that Equation (7) of Proposition 3.4 says us that for any vertical vector field ψ = ψ = ζ, the current ǫ (ζ d(hρ)) is a strong 1 2 ζ ⌋I L. Accornero and M. Palese 21 conserved current (i.e. conserved ‘off shell’). However, it can be easily checked that, at least in the specific case of study, for any (vertical) symmetry trans- formation of extremals ψ˜ = ψ the weak (i.e. ‘on shell’) conserved current reduces to ηρ(σηξ)ν(ψZ (ψBδ ) ψZ (ψBδ ))ds , which vanishes iden- ν ∇σ ρ BZ − ρ ∇σ ν BZ ξ tically because ηρ(σηξ)ν = ην(σηξ)ρ. This holds true for any couple of linearly dependent symmetry transformations.

Acknowledgements

Research partially supported by Department of Mathematics - University of Torino through the projects FERM RILO 17 01 and PALM RILO 20 01 (MP) and written under the auspices of GNSAGA-INdAM. The first author (LA) is also supported by a NWO-UGC project, Grant BM.00193.1. The second author (MP) would like to acknowledge the contribution of the Cost Action CA17139 - European Topology Interdisciplinary Action.

References

[1] L. Accornero: Jet prolongations and calculus of variations: second and higher order variations in the framework of the variational sequence, Master Thesis, University of Torino (2017).

[2] L. Accornero, M. Palese: The Jacobi morphism and the Hessian in higher order field theory; with applications to a Yang–Mills theory on a Minkowskian background, Int. J. Geom. Methods Mod. Phys. (2020) 2050114.

[3] M.F. Atiyah, R. Bott: The Yang–Mills equations over Riemann surfaces, Philos. Trans. Roy. Soc. London Ser. A 308 (1505) (1983) 523–615.

[4] E. Bessel-Hagen: Uber¨ die Erhaltungss¨atze der Elektrodynamik, Math. Ann. 84 (1921) 258–276.

[5] J.P. Bourguignon: Yang-Mills theory: the differential geometric side, Differential geometry (Lyngby, 1985), 13–54, Lecture Notes Math. 1263 Springer (Berlin, 1987).

[6] G.E. Bredon: Sheaf Theory, McGraw–Hill (New York, 1967). L. Accornero and M. Palese 22

[7] F. Cattafi, M. Palese, E. Winterroth: Variational derivatives in locally Lagrangian field theories and Noether–Bessel-Hagen currents, Int. J. Geom. Methods Mod. Phys. 13 (2016) 1650067, 16. [8] D.J. Eck: Gauge-natural bundles and generalized gauge theories, Mem. Amer. Math. Soc. 247 (1981) 1–48. [9] L. Fatibene, M. Francaviglia, M. Palese: Conservation laws and varia- tional sequences in gauge-natural theories, Math. Proc. Cambridge Phi- los. Soc. 130 (3) (2001) 555–569. [10] M. Ferraris, M. Francaviglia, M. Palese, E. Winterroth: Gauge-natural Noether currents and connection fields, Int. J. Geom. Methods Mod. Phys. 8(1) (2011) 177–185. [11] M. Ferraris, M. Palese, E. Winterroth: Local variational problems and conservation laws, Diff. Geom. Appl. 29 (1) (2011) S80–S85. [12] M. Francaviglia, M. Palese, R. Vitolo: The Hessian and Jacobi mor- phisms for higher order calculus of variations, Differential Geom. Appl.22(2005) 105–120 [13] M. Francaviglia, M. Palese, E. Winterroth: Variationally equivalent problems and variations of Noether currents, Int. J. Geom. Meth. Mod. Phys. 10(1) (2013) 1220024. [14] M. Francaviglia, M. Raiteri: Hamiltonian, energy and entropy in general relativity with non-orthogonal boundaries Class. Quantum Grav. 19 (2) (2002 ) 237. [15] H. Goldschmidt, S. Sternberg: The Hamilton-Cartan formalism in the calculus of variations, Ann. Inst. Fourier (Grenoble) 23 (1973) 203–267 [16] I. Kol´aˇr: Prolongations of generalized connections, Coll. Math. Soc. J´anos Bolyai, (Differential Geometry, Budapest, 1979) 31 (1979) 317– 325. [17] I. Kol´aˇr, P.W. Michor, J. Slov´ak: Natural Operations in Differential Geometry, (Springer–Verlag, N.Y., 1993). [18] Y. Kosmann-Schwarzbach: The Noether Theorems; translated from French by Bertram E. Schwarzbach (Springer 2011). L. Accornero and M. Palese 23

[19] M. Krbek, J. Musilov´a: Representation of the variational sequence by differential forms, Rep. Math. Phys. 51 (2-3) (2003) 251–258. [20] M. Krbek, J. Musilov´a: Representation of the variational sequence by differential forms, Acta Appl. Math. 88 (2) (2005) 177–199. [21] D. Krupka: Some geometric aspects of variational problems in fibred manifolds, Folia Fac. Sci. Nat. UJEP Brunensis 14, J. E. PurkynˇeUniv. (Brno, 1973) 1–65, arXiv: math-ph/0110005. [22] D. Krupka: Variational Sequences on Finite Order Jet Spaces, Proc. Diff. Geom. Appl.; J. Janyˇska, D. Krupka eds., World Sci. (Singapore, 1990) 236–254. [23] D. Krupka: Invariant variational structures on fibered manifolds Int. J. Geom. Methods Mod. Phys.12 (02) 1550020 (2015). [24] E. Noether: Invariante Variationsprobleme, Nachr. Ges. Wiss. G¨ott., Math. Phys. Kl. II (1918) 235–257. [25] M. Palese: Variations by generalized symmetries of local Noether strong currents equivalent to global canonical Noether currents, Comm. Math. 24 (2) (2016) pp. 125–135; Erratum, Comm. Math. 25 (1) (2017) 71–72. [26] M. Palese, O. Rossi, E. Winterroth, J. Musilov´a: Variational sequences, representation sequences and applications in physics, SIGMA 12 (2016) 045, 45 pages (2016). [27] M. Palese, O. Rossi, F. Zanello: Geometric integration by parts and Lepage equivalents, arXiv:2010.16135. [28] M. Palese, E. Winterroth: Global Generalized Bianchi Identities for Invariant Variational Problems on Gauge-natural Bundles, Arch. Math. (Brno) 41 (3) (2005) 289–310. [29] M. Palese, E. Winterroth: The relation between the Jacobi morphism and the Hessian in gauge-natural field theories, Theoret. Math. Phys. 152 (2) (2007) 1191–1200. [30] M. Palese, E. Winterroth: Topological obstructions in Lagrangian field theories, with an application to 3D Chern–Simons , J. Math. Phys 58 (2) (2017) 023502. L. Accornero and M. Palese 24

[31] M. Palese, E. Winterroth: Higgs fields induced by Yang-Mills type La- grangians on gauge-natural prolongations of principal bundles, Int. J. Geom. Meth. Mod. Phys. 16(3) (2019) 1950049, 15 pp.

[32] W. Sarlet, F. Cantrijn: Higher-order Noether symmetries and constants of the motion, J. Phys. A: Math. Gen. 14 (1981) 479–492.

[33] D.J. Saunders: The geometry of jet bundles, London Mathematical Soci- ety Lecture Note 142 Cambridge Univ. Press (Cambridge, 1989).

[34] G. ’t Hooft: Topology of the gauge condition and new confinement phases in non-abelian gauge theories, Nucl. Phys. B190[FS3] (1981) 455–478.

[35] A. Trautman: Noether equations and conservation laws, Comm. Math. Phys. 6 (1967) 248–261.

[36] C.N. Yang, R.L. Mills: Conservation of Isotopic Spin and Isotopic Gauge Invariance, Phys. Rev. 96 (1) (1954) 191–195.