NEW DEVELOPMENTS IN GEOMETRIC MECHANICS Janusz Grabowski (Polish - - PowerPoint PPT Presentation

new developments in geometric mechanics
SMART_READER_LITE
LIVE PREVIEW

NEW DEVELOPMENTS IN GEOMETRIC MECHANICS Janusz Grabowski (Polish - - PowerPoint PPT Presentation

NEW DEVELOPMENTS IN GEOMETRIC MECHANICS Janusz Grabowski (Polish Academy of Sciences) GEOMETRY OF JETS AND FIELDS B edlewo, 10-16 May, 2015 J.Grabowski (IMPAN) B edlewo, 10-16/05/2015 1 / 27 New developments in geometric mechanics


slide-1
SLIDE 1

NEW DEVELOPMENTS IN GEOMETRIC MECHANICS

Janusz Grabowski

(Polish Academy of Sciences)

GEOMETRY OF JETS AND FIELDS B¸ edlewo, 10-16 May, 2015

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 1 / 27

slide-2
SLIDE 2

Contents

Graded and double graded bundles Tulczyjew triples Mechanics on algebroids with vakonomic constraints Higher order Lagrangians Lagrangian framework for graded bundles Higher order Lagrangian mechanics on Lie algebroids Geometric mechanics of strings (optionally) The talk is based on some ideas of W. M. Tulczyjew and my collaboration with A. Bruce, K. Grabowska, M. Rotkiewicz and P. Urba´ nski: Grabowski-Rotkiewicz, J. Geom. Phys. 62 (2012), 21–36. Grabowska-Grabowski-Urba´ nski, J. Geom. Mech. 6 (2014), 503–526. Bruce-Grabowska-Grabowski, J. Phys. A 48 (2015), 205203 (32pp). Bruce-Grabowska-Grabowski, arXiv:1409.0439.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 2 / 27

slide-3
SLIDE 3

Contents

Graded and double graded bundles Tulczyjew triples Mechanics on algebroids with vakonomic constraints Higher order Lagrangians Lagrangian framework for graded bundles Higher order Lagrangian mechanics on Lie algebroids Geometric mechanics of strings (optionally) The talk is based on some ideas of W. M. Tulczyjew and my collaboration with A. Bruce, K. Grabowska, M. Rotkiewicz and P. Urba´ nski: Grabowski-Rotkiewicz, J. Geom. Phys. 62 (2012), 21–36. Grabowska-Grabowski-Urba´ nski, J. Geom. Mech. 6 (2014), 503–526. Bruce-Grabowska-Grabowski, J. Phys. A 48 (2015), 205203 (32pp). Bruce-Grabowska-Grabowski, arXiv:1409.0439.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 2 / 27

slide-4
SLIDE 4

Contents

Graded and double graded bundles Tulczyjew triples Mechanics on algebroids with vakonomic constraints Higher order Lagrangians Lagrangian framework for graded bundles Higher order Lagrangian mechanics on Lie algebroids Geometric mechanics of strings (optionally) The talk is based on some ideas of W. M. Tulczyjew and my collaboration with A. Bruce, K. Grabowska, M. Rotkiewicz and P. Urba´ nski: Grabowski-Rotkiewicz, J. Geom. Phys. 62 (2012), 21–36. Grabowska-Grabowski-Urba´ nski, J. Geom. Mech. 6 (2014), 503–526. Bruce-Grabowska-Grabowski, J. Phys. A 48 (2015), 205203 (32pp). Bruce-Grabowska-Grabowski, arXiv:1409.0439.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 2 / 27

slide-5
SLIDE 5

Contents

Graded and double graded bundles Tulczyjew triples Mechanics on algebroids with vakonomic constraints Higher order Lagrangians Lagrangian framework for graded bundles Higher order Lagrangian mechanics on Lie algebroids Geometric mechanics of strings (optionally) The talk is based on some ideas of W. M. Tulczyjew and my collaboration with A. Bruce, K. Grabowska, M. Rotkiewicz and P. Urba´ nski: Grabowski-Rotkiewicz, J. Geom. Phys. 62 (2012), 21–36. Grabowska-Grabowski-Urba´ nski, J. Geom. Mech. 6 (2014), 503–526. Bruce-Grabowska-Grabowski, J. Phys. A 48 (2015), 205203 (32pp). Bruce-Grabowska-Grabowski, arXiv:1409.0439.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 2 / 27

slide-6
SLIDE 6

Contents

Graded and double graded bundles Tulczyjew triples Mechanics on algebroids with vakonomic constraints Higher order Lagrangians Lagrangian framework for graded bundles Higher order Lagrangian mechanics on Lie algebroids Geometric mechanics of strings (optionally) The talk is based on some ideas of W. M. Tulczyjew and my collaboration with A. Bruce, K. Grabowska, M. Rotkiewicz and P. Urba´ nski: Grabowski-Rotkiewicz, J. Geom. Phys. 62 (2012), 21–36. Grabowska-Grabowski-Urba´ nski, J. Geom. Mech. 6 (2014), 503–526. Bruce-Grabowska-Grabowski, J. Phys. A 48 (2015), 205203 (32pp). Bruce-Grabowska-Grabowski, arXiv:1409.0439.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 2 / 27

slide-7
SLIDE 7

Contents

Graded and double graded bundles Tulczyjew triples Mechanics on algebroids with vakonomic constraints Higher order Lagrangians Lagrangian framework for graded bundles Higher order Lagrangian mechanics on Lie algebroids Geometric mechanics of strings (optionally) The talk is based on some ideas of W. M. Tulczyjew and my collaboration with A. Bruce, K. Grabowska, M. Rotkiewicz and P. Urba´ nski: Grabowski-Rotkiewicz, J. Geom. Phys. 62 (2012), 21–36. Grabowska-Grabowski-Urba´ nski, J. Geom. Mech. 6 (2014), 503–526. Bruce-Grabowska-Grabowski, J. Phys. A 48 (2015), 205203 (32pp). Bruce-Grabowska-Grabowski, arXiv:1409.0439.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 2 / 27

slide-8
SLIDE 8

Contents

Graded and double graded bundles Tulczyjew triples Mechanics on algebroids with vakonomic constraints Higher order Lagrangians Lagrangian framework for graded bundles Higher order Lagrangian mechanics on Lie algebroids Geometric mechanics of strings (optionally) The talk is based on some ideas of W. M. Tulczyjew and my collaboration with A. Bruce, K. Grabowska, M. Rotkiewicz and P. Urba´ nski: Grabowski-Rotkiewicz, J. Geom. Phys. 62 (2012), 21–36. Grabowska-Grabowski-Urba´ nski, J. Geom. Mech. 6 (2014), 503–526. Bruce-Grabowska-Grabowski, J. Phys. A 48 (2015), 205203 (32pp). Bruce-Grabowska-Grabowski, arXiv:1409.0439.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 2 / 27

slide-9
SLIDE 9

Contents

Graded and double graded bundles Tulczyjew triples Mechanics on algebroids with vakonomic constraints Higher order Lagrangians Lagrangian framework for graded bundles Higher order Lagrangian mechanics on Lie algebroids Geometric mechanics of strings (optionally) The talk is based on some ideas of W. M. Tulczyjew and my collaboration with A. Bruce, K. Grabowska, M. Rotkiewicz and P. Urba´ nski: Grabowski-Rotkiewicz, J. Geom. Phys. 62 (2012), 21–36. Grabowska-Grabowski-Urba´ nski, J. Geom. Mech. 6 (2014), 503–526. Bruce-Grabowska-Grabowski, J. Phys. A 48 (2015), 205203 (32pp). Bruce-Grabowska-Grabowski, arXiv:1409.0439.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 2 / 27

slide-10
SLIDE 10

Contents

Graded and double graded bundles Tulczyjew triples Mechanics on algebroids with vakonomic constraints Higher order Lagrangians Lagrangian framework for graded bundles Higher order Lagrangian mechanics on Lie algebroids Geometric mechanics of strings (optionally) The talk is based on some ideas of W. M. Tulczyjew and my collaboration with A. Bruce, K. Grabowska, M. Rotkiewicz and P. Urba´ nski: Grabowski-Rotkiewicz, J. Geom. Phys. 62 (2012), 21–36. Grabowska-Grabowski-Urba´ nski, J. Geom. Mech. 6 (2014), 503–526. Bruce-Grabowska-Grabowski, J. Phys. A 48 (2015), 205203 (32pp). Bruce-Grabowska-Grabowski, arXiv:1409.0439.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 2 / 27

slide-11
SLIDE 11

Contents

Graded and double graded bundles Tulczyjew triples Mechanics on algebroids with vakonomic constraints Higher order Lagrangians Lagrangian framework for graded bundles Higher order Lagrangian mechanics on Lie algebroids Geometric mechanics of strings (optionally) The talk is based on some ideas of W. M. Tulczyjew and my collaboration with A. Bruce, K. Grabowska, M. Rotkiewicz and P. Urba´ nski: Grabowski-Rotkiewicz, J. Geom. Phys. 62 (2012), 21–36. Grabowska-Grabowski-Urba´ nski, J. Geom. Mech. 6 (2014), 503–526. Bruce-Grabowska-Grabowski, J. Phys. A 48 (2015), 205203 (32pp). Bruce-Grabowska-Grabowski, arXiv:1409.0439.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 2 / 27

slide-12
SLIDE 12

Contents

Graded and double graded bundles Tulczyjew triples Mechanics on algebroids with vakonomic constraints Higher order Lagrangians Lagrangian framework for graded bundles Higher order Lagrangian mechanics on Lie algebroids Geometric mechanics of strings (optionally) The talk is based on some ideas of W. M. Tulczyjew and my collaboration with A. Bruce, K. Grabowska, M. Rotkiewicz and P. Urba´ nski: Grabowski-Rotkiewicz, J. Geom. Phys. 62 (2012), 21–36. Grabowska-Grabowski-Urba´ nski, J. Geom. Mech. 6 (2014), 503–526. Bruce-Grabowska-Grabowski, J. Phys. A 48 (2015), 205203 (32pp). Bruce-Grabowska-Grabowski, arXiv:1409.0439.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 2 / 27

slide-13
SLIDE 13

Contents

Graded and double graded bundles Tulczyjew triples Mechanics on algebroids with vakonomic constraints Higher order Lagrangians Lagrangian framework for graded bundles Higher order Lagrangian mechanics on Lie algebroids Geometric mechanics of strings (optionally) The talk is based on some ideas of W. M. Tulczyjew and my collaboration with A. Bruce, K. Grabowska, M. Rotkiewicz and P. Urba´ nski: Grabowski-Rotkiewicz, J. Geom. Phys. 62 (2012), 21–36. Grabowska-Grabowski-Urba´ nski, J. Geom. Mech. 6 (2014), 503–526. Bruce-Grabowska-Grabowski, J. Phys. A 48 (2015), 205203 (32pp). Bruce-Grabowska-Grabowski, arXiv:1409.0439.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 2 / 27

slide-14
SLIDE 14

Contents

Graded and double graded bundles Tulczyjew triples Mechanics on algebroids with vakonomic constraints Higher order Lagrangians Lagrangian framework for graded bundles Higher order Lagrangian mechanics on Lie algebroids Geometric mechanics of strings (optionally) The talk is based on some ideas of W. M. Tulczyjew and my collaboration with A. Bruce, K. Grabowska, M. Rotkiewicz and P. Urba´ nski: Grabowski-Rotkiewicz, J. Geom. Phys. 62 (2012), 21–36. Grabowska-Grabowski-Urba´ nski, J. Geom. Mech. 6 (2014), 503–526. Bruce-Grabowska-Grabowski, J. Phys. A 48 (2015), 205203 (32pp). Bruce-Grabowska-Grabowski, arXiv:1409.0439.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 2 / 27

slide-15
SLIDE 15

Vector bundles as graded bundles

A vector bundle is a locally trivial fibration τ : E → M which, locally

  • ver U ⊂ M, reads τ −1(U) ≃ U × Rn and admits an atlas in which

local trivializations transform linearly in fibers U ∩ V × Rn ∋ (x, y) → (x, A(x)y) ∈ U ∩ V × Rn , A(x) ∈ GL(n, R). The latter property can also be expressed in the terms of the gradation in which base coordinates x have degrees 0 and ‘linear coordinates’ y have degree 1. Linearity in y′s is now equivalent to the fact that changes of coordinates respect the degrees. Morphisms in the category of vector bundles are represented by commutative diagram of smooth maps E1

Φ

  • τ1
  • E2

τ2

  • M1

ϕ

M2

being linear in fibres.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 3 / 27

slide-16
SLIDE 16

Vector bundles as graded bundles

A vector bundle is a locally trivial fibration τ : E → M which, locally

  • ver U ⊂ M, reads τ −1(U) ≃ U × Rn and admits an atlas in which

local trivializations transform linearly in fibers U ∩ V × Rn ∋ (x, y) → (x, A(x)y) ∈ U ∩ V × Rn , A(x) ∈ GL(n, R). The latter property can also be expressed in the terms of the gradation in which base coordinates x have degrees 0 and ‘linear coordinates’ y have degree 1. Linearity in y′s is now equivalent to the fact that changes of coordinates respect the degrees. Morphisms in the category of vector bundles are represented by commutative diagram of smooth maps E1

Φ

  • τ1
  • E2

τ2

  • M1

ϕ

M2

being linear in fibres.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 3 / 27

slide-17
SLIDE 17

Vector bundles as graded bundles

A vector bundle is a locally trivial fibration τ : E → M which, locally

  • ver U ⊂ M, reads τ −1(U) ≃ U × Rn and admits an atlas in which

local trivializations transform linearly in fibers U ∩ V × Rn ∋ (x, y) → (x, A(x)y) ∈ U ∩ V × Rn , A(x) ∈ GL(n, R). The latter property can also be expressed in the terms of the gradation in which base coordinates x have degrees 0 and ‘linear coordinates’ y have degree 1. Linearity in y′s is now equivalent to the fact that changes of coordinates respect the degrees. Morphisms in the category of vector bundles are represented by commutative diagram of smooth maps E1

Φ

  • τ1
  • E2

τ2

  • M1

ϕ

M2

being linear in fibres.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 3 / 27

slide-18
SLIDE 18

Vector bundles as graded bundles

A vector bundle is a locally trivial fibration τ : E → M which, locally

  • ver U ⊂ M, reads τ −1(U) ≃ U × Rn and admits an atlas in which

local trivializations transform linearly in fibers U ∩ V × Rn ∋ (x, y) → (x, A(x)y) ∈ U ∩ V × Rn , A(x) ∈ GL(n, R). The latter property can also be expressed in the terms of the gradation in which base coordinates x have degrees 0 and ‘linear coordinates’ y have degree 1. Linearity in y′s is now equivalent to the fact that changes of coordinates respect the degrees. Morphisms in the category of vector bundles are represented by commutative diagram of smooth maps E1

Φ

  • τ1
  • E2

τ2

  • M1

ϕ

M2

being linear in fibres.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 3 / 27

slide-19
SLIDE 19

Vector bundles as graded bundles

A vector bundle is a locally trivial fibration τ : E → M which, locally

  • ver U ⊂ M, reads τ −1(U) ≃ U × Rn and admits an atlas in which

local trivializations transform linearly in fibers U ∩ V × Rn ∋ (x, y) → (x, A(x)y) ∈ U ∩ V × Rn , A(x) ∈ GL(n, R). The latter property can also be expressed in the terms of the gradation in which base coordinates x have degrees 0 and ‘linear coordinates’ y have degree 1. Linearity in y′s is now equivalent to the fact that changes of coordinates respect the degrees. Morphisms in the category of vector bundles are represented by commutative diagram of smooth maps E1

Φ

  • τ1
  • E2

τ2

  • M1

ϕ

M2

being linear in fibres.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 3 / 27

slide-20
SLIDE 20

Vector bundles as graded bundles

A vector bundle is a locally trivial fibration τ : E → M which, locally

  • ver U ⊂ M, reads τ −1(U) ≃ U × Rn and admits an atlas in which

local trivializations transform linearly in fibers U ∩ V × Rn ∋ (x, y) → (x, A(x)y) ∈ U ∩ V × Rn , A(x) ∈ GL(n, R). The latter property can also be expressed in the terms of the gradation in which base coordinates x have degrees 0 and ‘linear coordinates’ y have degree 1. Linearity in y′s is now equivalent to the fact that changes of coordinates respect the degrees. Morphisms in the category of vector bundles are represented by commutative diagram of smooth maps E1

Φ

  • τ1
  • E2

τ2

  • M1

ϕ

M2

being linear in fibres.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 3 / 27

slide-21
SLIDE 21

Vector bundles as graded bundles

A vector bundle is a locally trivial fibration τ : E → M which, locally

  • ver U ⊂ M, reads τ −1(U) ≃ U × Rn and admits an atlas in which

local trivializations transform linearly in fibers U ∩ V × Rn ∋ (x, y) → (x, A(x)y) ∈ U ∩ V × Rn , A(x) ∈ GL(n, R). The latter property can also be expressed in the terms of the gradation in which base coordinates x have degrees 0 and ‘linear coordinates’ y have degree 1. Linearity in y′s is now equivalent to the fact that changes of coordinates respect the degrees. Morphisms in the category of vector bundles are represented by commutative diagram of smooth maps E1

Φ

  • τ1
  • E2

τ2

  • M1

ϕ

M2

being linear in fibres.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 3 / 27

slide-22
SLIDE 22

Vector bundles as graded bundles

A vector bundle is a locally trivial fibration τ : E → M which, locally

  • ver U ⊂ M, reads τ −1(U) ≃ U × Rn and admits an atlas in which

local trivializations transform linearly in fibers U ∩ V × Rn ∋ (x, y) → (x, A(x)y) ∈ U ∩ V × Rn , A(x) ∈ GL(n, R). The latter property can also be expressed in the terms of the gradation in which base coordinates x have degrees 0 and ‘linear coordinates’ y have degree 1. Linearity in y′s is now equivalent to the fact that changes of coordinates respect the degrees. Morphisms in the category of vector bundles are represented by commutative diagram of smooth maps E1

Φ

  • τ1
  • E2

τ2

  • M1

ϕ

M2

being linear in fibres.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 3 / 27

slide-23
SLIDE 23

Graded bundles

Canonical examples and constructions: TM, T∗M, E ⊗M F, ∧kE, etc. A straightforward generalization is the concept of a graded bundle τ : F → M with a local trivialization by U × Rn as before, and with the difference that the local coordinates (y1, . . . , yn) in the fibres have now associated positive integer weights w1, . . . , wn, that are preserved by changes of local trivializations: U ∩ V × Rn ∋ (x, y) → (x, A(x, y)) ∈ U ∩ V × Rn , One can show that in this case A(x, y) must be polynomial in fiber coordinates, i.e. any graded bundle is a polynomial bundle. As these polynomials need not to be linear, graded bundles do not have, in general, vector space structure in fibers. For instance, if (y, z) ∈ R2 are coordinates of degrees 1, 2, respectively, then the map (y, z) → (y, z + y2) is a diffeomorphism preserving the degrees, but it is nonlinear. If all wi ≤ r, we say that the graded bundle is of degree r.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 4 / 27

slide-24
SLIDE 24

Graded bundles

Canonical examples and constructions: TM, T∗M, E ⊗M F, ∧kE, etc. A straightforward generalization is the concept of a graded bundle τ : F → M with a local trivialization by U × Rn as before, and with the difference that the local coordinates (y1, . . . , yn) in the fibres have now associated positive integer weights w1, . . . , wn, that are preserved by changes of local trivializations: U ∩ V × Rn ∋ (x, y) → (x, A(x, y)) ∈ U ∩ V × Rn , One can show that in this case A(x, y) must be polynomial in fiber coordinates, i.e. any graded bundle is a polynomial bundle. As these polynomials need not to be linear, graded bundles do not have, in general, vector space structure in fibers. For instance, if (y, z) ∈ R2 are coordinates of degrees 1, 2, respectively, then the map (y, z) → (y, z + y2) is a diffeomorphism preserving the degrees, but it is nonlinear. If all wi ≤ r, we say that the graded bundle is of degree r.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 4 / 27

slide-25
SLIDE 25

Graded bundles

Canonical examples and constructions: TM, T∗M, E ⊗M F, ∧kE, etc. A straightforward generalization is the concept of a graded bundle τ : F → M with a local trivialization by U × Rn as before, and with the difference that the local coordinates (y1, . . . , yn) in the fibres have now associated positive integer weights w1, . . . , wn, that are preserved by changes of local trivializations: U ∩ V × Rn ∋ (x, y) → (x, A(x, y)) ∈ U ∩ V × Rn , One can show that in this case A(x, y) must be polynomial in fiber coordinates, i.e. any graded bundle is a polynomial bundle. As these polynomials need not to be linear, graded bundles do not have, in general, vector space structure in fibers. For instance, if (y, z) ∈ R2 are coordinates of degrees 1, 2, respectively, then the map (y, z) → (y, z + y2) is a diffeomorphism preserving the degrees, but it is nonlinear. If all wi ≤ r, we say that the graded bundle is of degree r.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 4 / 27

slide-26
SLIDE 26

Graded bundles

Canonical examples and constructions: TM, T∗M, E ⊗M F, ∧kE, etc. A straightforward generalization is the concept of a graded bundle τ : F → M with a local trivialization by U × Rn as before, and with the difference that the local coordinates (y1, . . . , yn) in the fibres have now associated positive integer weights w1, . . . , wn, that are preserved by changes of local trivializations: U ∩ V × Rn ∋ (x, y) → (x, A(x, y)) ∈ U ∩ V × Rn , One can show that in this case A(x, y) must be polynomial in fiber coordinates, i.e. any graded bundle is a polynomial bundle. As these polynomials need not to be linear, graded bundles do not have, in general, vector space structure in fibers. For instance, if (y, z) ∈ R2 are coordinates of degrees 1, 2, respectively, then the map (y, z) → (y, z + y2) is a diffeomorphism preserving the degrees, but it is nonlinear. If all wi ≤ r, we say that the graded bundle is of degree r.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 4 / 27

slide-27
SLIDE 27

Graded bundles

Canonical examples and constructions: TM, T∗M, E ⊗M F, ∧kE, etc. A straightforward generalization is the concept of a graded bundle τ : F → M with a local trivialization by U × Rn as before, and with the difference that the local coordinates (y1, . . . , yn) in the fibres have now associated positive integer weights w1, . . . , wn, that are preserved by changes of local trivializations: U ∩ V × Rn ∋ (x, y) → (x, A(x, y)) ∈ U ∩ V × Rn , One can show that in this case A(x, y) must be polynomial in fiber coordinates, i.e. any graded bundle is a polynomial bundle. As these polynomials need not to be linear, graded bundles do not have, in general, vector space structure in fibers. For instance, if (y, z) ∈ R2 are coordinates of degrees 1, 2, respectively, then the map (y, z) → (y, z + y2) is a diffeomorphism preserving the degrees, but it is nonlinear. If all wi ≤ r, we say that the graded bundle is of degree r.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 4 / 27

slide-28
SLIDE 28

Graded bundles

Canonical examples and constructions: TM, T∗M, E ⊗M F, ∧kE, etc. A straightforward generalization is the concept of a graded bundle τ : F → M with a local trivialization by U × Rn as before, and with the difference that the local coordinates (y1, . . . , yn) in the fibres have now associated positive integer weights w1, . . . , wn, that are preserved by changes of local trivializations: U ∩ V × Rn ∋ (x, y) → (x, A(x, y)) ∈ U ∩ V × Rn , One can show that in this case A(x, y) must be polynomial in fiber coordinates, i.e. any graded bundle is a polynomial bundle. As these polynomials need not to be linear, graded bundles do not have, in general, vector space structure in fibers. For instance, if (y, z) ∈ R2 are coordinates of degrees 1, 2, respectively, then the map (y, z) → (y, z + y2) is a diffeomorphism preserving the degrees, but it is nonlinear. If all wi ≤ r, we say that the graded bundle is of degree r.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 4 / 27

slide-29
SLIDE 29

Graded bundles

Canonical examples and constructions: TM, T∗M, E ⊗M F, ∧kE, etc. A straightforward generalization is the concept of a graded bundle τ : F → M with a local trivialization by U × Rn as before, and with the difference that the local coordinates (y1, . . . , yn) in the fibres have now associated positive integer weights w1, . . . , wn, that are preserved by changes of local trivializations: U ∩ V × Rn ∋ (x, y) → (x, A(x, y)) ∈ U ∩ V × Rn , One can show that in this case A(x, y) must be polynomial in fiber coordinates, i.e. any graded bundle is a polynomial bundle. As these polynomials need not to be linear, graded bundles do not have, in general, vector space structure in fibers. For instance, if (y, z) ∈ R2 are coordinates of degrees 1, 2, respectively, then the map (y, z) → (y, z + y2) is a diffeomorphism preserving the degrees, but it is nonlinear. If all wi ≤ r, we say that the graded bundle is of degree r.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 4 / 27

slide-30
SLIDE 30

Graded bundles

Canonical examples and constructions: TM, T∗M, E ⊗M F, ∧kE, etc. A straightforward generalization is the concept of a graded bundle τ : F → M with a local trivialization by U × Rn as before, and with the difference that the local coordinates (y1, . . . , yn) in the fibres have now associated positive integer weights w1, . . . , wn, that are preserved by changes of local trivializations: U ∩ V × Rn ∋ (x, y) → (x, A(x, y)) ∈ U ∩ V × Rn , One can show that in this case A(x, y) must be polynomial in fiber coordinates, i.e. any graded bundle is a polynomial bundle. As these polynomials need not to be linear, graded bundles do not have, in general, vector space structure in fibers. For instance, if (y, z) ∈ R2 are coordinates of degrees 1, 2, respectively, then the map (y, z) → (y, z + y2) is a diffeomorphism preserving the degrees, but it is nonlinear. If all wi ≤ r, we say that the graded bundle is of degree r.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 4 / 27

slide-31
SLIDE 31

Graded bundles

Canonical examples and constructions: TM, T∗M, E ⊗M F, ∧kE, etc. A straightforward generalization is the concept of a graded bundle τ : F → M with a local trivialization by U × Rn as before, and with the difference that the local coordinates (y1, . . . , yn) in the fibres have now associated positive integer weights w1, . . . , wn, that are preserved by changes of local trivializations: U ∩ V × Rn ∋ (x, y) → (x, A(x, y)) ∈ U ∩ V × Rn , One can show that in this case A(x, y) must be polynomial in fiber coordinates, i.e. any graded bundle is a polynomial bundle. As these polynomials need not to be linear, graded bundles do not have, in general, vector space structure in fibers. For instance, if (y, z) ∈ R2 are coordinates of degrees 1, 2, respectively, then the map (y, z) → (y, z + y2) is a diffeomorphism preserving the degrees, but it is nonlinear. If all wi ≤ r, we say that the graded bundle is of degree r.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 4 / 27

slide-32
SLIDE 32

Graded bundles

In the above terminology, vector bundles are just graded bundles of degree 1. Graded bundles Fk of degree k admit, like many jet bundles, a tower

  • f affine fibrations by their subbundles of lower degrees

Fk

τ k

− → Fk−1

τ k−1

− → · · ·

τ 3

− → F2

τ 2

− → F1

τ 1

− → F0 = M . Canonical examples: TkM, with canonical coordinates (x, ˙ x, ¨ x, ... x , . . . )

  • f degrees, respectively, 0, 1, 2, 3, etc.

Another example. If τ : E → M is a vector bundle, then ∧rTE is canonically a graded bundle of degree r with respect to the projection ∧rTτ : ∧rTE → ∧rTM . Note that similar objects has been used in supergeometry by Kosmann-Schwarzbach, Voronov, Mackenzie, Roytenberg et al. under the name N-manifolds. However, we will work with classical, purely even manifolds during this talk.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 5 / 27

slide-33
SLIDE 33

Graded bundles

In the above terminology, vector bundles are just graded bundles of degree 1. Graded bundles Fk of degree k admit, like many jet bundles, a tower

  • f affine fibrations by their subbundles of lower degrees

Fk

τ k

− → Fk−1

τ k−1

− → · · ·

τ 3

− → F2

τ 2

− → F1

τ 1

− → F0 = M . Canonical examples: TkM, with canonical coordinates (x, ˙ x, ¨ x, ... x , . . . )

  • f degrees, respectively, 0, 1, 2, 3, etc.

Another example. If τ : E → M is a vector bundle, then ∧rTE is canonically a graded bundle of degree r with respect to the projection ∧rTτ : ∧rTE → ∧rTM . Note that similar objects has been used in supergeometry by Kosmann-Schwarzbach, Voronov, Mackenzie, Roytenberg et al. under the name N-manifolds. However, we will work with classical, purely even manifolds during this talk.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 5 / 27

slide-34
SLIDE 34

Graded bundles

In the above terminology, vector bundles are just graded bundles of degree 1. Graded bundles Fk of degree k admit, like many jet bundles, a tower

  • f affine fibrations by their subbundles of lower degrees

Fk

τ k

− → Fk−1

τ k−1

− → · · ·

τ 3

− → F2

τ 2

− → F1

τ 1

− → F0 = M . Canonical examples: TkM, with canonical coordinates (x, ˙ x, ¨ x, ... x , . . . )

  • f degrees, respectively, 0, 1, 2, 3, etc.

Another example. If τ : E → M is a vector bundle, then ∧rTE is canonically a graded bundle of degree r with respect to the projection ∧rTτ : ∧rTE → ∧rTM . Note that similar objects has been used in supergeometry by Kosmann-Schwarzbach, Voronov, Mackenzie, Roytenberg et al. under the name N-manifolds. However, we will work with classical, purely even manifolds during this talk.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 5 / 27

slide-35
SLIDE 35

Graded bundles

In the above terminology, vector bundles are just graded bundles of degree 1. Graded bundles Fk of degree k admit, like many jet bundles, a tower

  • f affine fibrations by their subbundles of lower degrees

Fk

τ k

− → Fk−1

τ k−1

− → · · ·

τ 3

− → F2

τ 2

− → F1

τ 1

− → F0 = M . Canonical examples: TkM, with canonical coordinates (x, ˙ x, ¨ x, ... x , . . . )

  • f degrees, respectively, 0, 1, 2, 3, etc.

Another example. If τ : E → M is a vector bundle, then ∧rTE is canonically a graded bundle of degree r with respect to the projection ∧rTτ : ∧rTE → ∧rTM . Note that similar objects has been used in supergeometry by Kosmann-Schwarzbach, Voronov, Mackenzie, Roytenberg et al. under the name N-manifolds. However, we will work with classical, purely even manifolds during this talk.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 5 / 27

slide-36
SLIDE 36

Graded bundles

In the above terminology, vector bundles are just graded bundles of degree 1. Graded bundles Fk of degree k admit, like many jet bundles, a tower

  • f affine fibrations by their subbundles of lower degrees

Fk

τ k

− → Fk−1

τ k−1

− → · · ·

τ 3

− → F2

τ 2

− → F1

τ 1

− → F0 = M . Canonical examples: TkM, with canonical coordinates (x, ˙ x, ¨ x, ... x , . . . )

  • f degrees, respectively, 0, 1, 2, 3, etc.

Another example. If τ : E → M is a vector bundle, then ∧rTE is canonically a graded bundle of degree r with respect to the projection ∧rTτ : ∧rTE → ∧rTM . Note that similar objects has been used in supergeometry by Kosmann-Schwarzbach, Voronov, Mackenzie, Roytenberg et al. under the name N-manifolds. However, we will work with classical, purely even manifolds during this talk.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 5 / 27

slide-37
SLIDE 37

Graded bundles

In the above terminology, vector bundles are just graded bundles of degree 1. Graded bundles Fk of degree k admit, like many jet bundles, a tower

  • f affine fibrations by their subbundles of lower degrees

Fk

τ k

− → Fk−1

τ k−1

− → · · ·

τ 3

− → F2

τ 2

− → F1

τ 1

− → F0 = M . Canonical examples: TkM, with canonical coordinates (x, ˙ x, ¨ x, ... x , . . . )

  • f degrees, respectively, 0, 1, 2, 3, etc.

Another example. If τ : E → M is a vector bundle, then ∧rTE is canonically a graded bundle of degree r with respect to the projection ∧rTτ : ∧rTE → ∧rTM . Note that similar objects has been used in supergeometry by Kosmann-Schwarzbach, Voronov, Mackenzie, Roytenberg et al. under the name N-manifolds. However, we will work with classical, purely even manifolds during this talk.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 5 / 27

slide-38
SLIDE 38

Graded bundles

In the above terminology, vector bundles are just graded bundles of degree 1. Graded bundles Fk of degree k admit, like many jet bundles, a tower

  • f affine fibrations by their subbundles of lower degrees

Fk

τ k

− → Fk−1

τ k−1

− → · · ·

τ 3

− → F2

τ 2

− → F1

τ 1

− → F0 = M . Canonical examples: TkM, with canonical coordinates (x, ˙ x, ¨ x, ... x , . . . )

  • f degrees, respectively, 0, 1, 2, 3, etc.

Another example. If τ : E → M is a vector bundle, then ∧rTE is canonically a graded bundle of degree r with respect to the projection ∧rTτ : ∧rTE → ∧rTM . Note that similar objects has been used in supergeometry by Kosmann-Schwarzbach, Voronov, Mackenzie, Roytenberg et al. under the name N-manifolds. However, we will work with classical, purely even manifolds during this talk.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 5 / 27

slide-39
SLIDE 39

Graded Bundles

With the use of coordinates (xα, ya) with degrees 0 for basic coordinates xα, and degrees wa > 0 for the fibre coordinates ya, we can define on the graded bundle F a globally defined weight vector field (Euler vector field) ∇F =

  • a

waya∂ya . The flow of the weight vector field extends to a smooth action R ∋ t → ht of multiplicative reals on F, ht(xµ, ya) = (xµ, twaya). Such an action h : R × F → F, ht ◦ hs = hts, we will call a homogeneity structure. A function f : F → R is called homogeneous of degree (weight) k if f (ht(x)) = tkf (x); similarly for the homogeneity of tensor fields. Morphisms of two homogeneity structures (Fi, hi), i = 1, 2, are defined as smooth maps Φ : F1 → F2 intertwining the R-actions: Φ ◦ h1

t = h2 t ◦ Φ. Consequently, a homogeneity substructure is a

smooth submanifold S invariant with respect to h, ht(S) ⊂ S.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 6 / 27

slide-40
SLIDE 40

Graded Bundles

With the use of coordinates (xα, ya) with degrees 0 for basic coordinates xα, and degrees wa > 0 for the fibre coordinates ya, we can define on the graded bundle F a globally defined weight vector field (Euler vector field) ∇F =

  • a

waya∂ya . The flow of the weight vector field extends to a smooth action R ∋ t → ht of multiplicative reals on F, ht(xµ, ya) = (xµ, twaya). Such an action h : R × F → F, ht ◦ hs = hts, we will call a homogeneity structure. A function f : F → R is called homogeneous of degree (weight) k if f (ht(x)) = tkf (x); similarly for the homogeneity of tensor fields. Morphisms of two homogeneity structures (Fi, hi), i = 1, 2, are defined as smooth maps Φ : F1 → F2 intertwining the R-actions: Φ ◦ h1

t = h2 t ◦ Φ. Consequently, a homogeneity substructure is a

smooth submanifold S invariant with respect to h, ht(S) ⊂ S.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 6 / 27

slide-41
SLIDE 41

Graded Bundles

With the use of coordinates (xα, ya) with degrees 0 for basic coordinates xα, and degrees wa > 0 for the fibre coordinates ya, we can define on the graded bundle F a globally defined weight vector field (Euler vector field) ∇F =

  • a

waya∂ya . The flow of the weight vector field extends to a smooth action R ∋ t → ht of multiplicative reals on F, ht(xµ, ya) = (xµ, twaya). Such an action h : R × F → F, ht ◦ hs = hts, we will call a homogeneity structure. A function f : F → R is called homogeneous of degree (weight) k if f (ht(x)) = tkf (x); similarly for the homogeneity of tensor fields. Morphisms of two homogeneity structures (Fi, hi), i = 1, 2, are defined as smooth maps Φ : F1 → F2 intertwining the R-actions: Φ ◦ h1

t = h2 t ◦ Φ. Consequently, a homogeneity substructure is a

smooth submanifold S invariant with respect to h, ht(S) ⊂ S.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 6 / 27

slide-42
SLIDE 42

Graded Bundles

With the use of coordinates (xα, ya) with degrees 0 for basic coordinates xα, and degrees wa > 0 for the fibre coordinates ya, we can define on the graded bundle F a globally defined weight vector field (Euler vector field) ∇F =

  • a

waya∂ya . The flow of the weight vector field extends to a smooth action R ∋ t → ht of multiplicative reals on F, ht(xµ, ya) = (xµ, twaya). Such an action h : R × F → F, ht ◦ hs = hts, we will call a homogeneity structure. A function f : F → R is called homogeneous of degree (weight) k if f (ht(x)) = tkf (x); similarly for the homogeneity of tensor fields. Morphisms of two homogeneity structures (Fi, hi), i = 1, 2, are defined as smooth maps Φ : F1 → F2 intertwining the R-actions: Φ ◦ h1

t = h2 t ◦ Φ. Consequently, a homogeneity substructure is a

smooth submanifold S invariant with respect to h, ht(S) ⊂ S.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 6 / 27

slide-43
SLIDE 43

Graded Bundles

With the use of coordinates (xα, ya) with degrees 0 for basic coordinates xα, and degrees wa > 0 for the fibre coordinates ya, we can define on the graded bundle F a globally defined weight vector field (Euler vector field) ∇F =

  • a

waya∂ya . The flow of the weight vector field extends to a smooth action R ∋ t → ht of multiplicative reals on F, ht(xµ, ya) = (xµ, twaya). Such an action h : R × F → F, ht ◦ hs = hts, we will call a homogeneity structure. A function f : F → R is called homogeneous of degree (weight) k if f (ht(x)) = tkf (x); similarly for the homogeneity of tensor fields. Morphisms of two homogeneity structures (Fi, hi), i = 1, 2, are defined as smooth maps Φ : F1 → F2 intertwining the R-actions: Φ ◦ h1

t = h2 t ◦ Φ. Consequently, a homogeneity substructure is a

smooth submanifold S invariant with respect to h, ht(S) ⊂ S.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 6 / 27

slide-44
SLIDE 44

Graded Bundles

With the use of coordinates (xα, ya) with degrees 0 for basic coordinates xα, and degrees wa > 0 for the fibre coordinates ya, we can define on the graded bundle F a globally defined weight vector field (Euler vector field) ∇F =

  • a

waya∂ya . The flow of the weight vector field extends to a smooth action R ∋ t → ht of multiplicative reals on F, ht(xµ, ya) = (xµ, twaya). Such an action h : R × F → F, ht ◦ hs = hts, we will call a homogeneity structure. A function f : F → R is called homogeneous of degree (weight) k if f (ht(x)) = tkf (x); similarly for the homogeneity of tensor fields. Morphisms of two homogeneity structures (Fi, hi), i = 1, 2, are defined as smooth maps Φ : F1 → F2 intertwining the R-actions: Φ ◦ h1

t = h2 t ◦ Φ. Consequently, a homogeneity substructure is a

smooth submanifold S invariant with respect to h, ht(S) ⊂ S.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 6 / 27

slide-45
SLIDE 45

Graded Bundles

With the use of coordinates (xα, ya) with degrees 0 for basic coordinates xα, and degrees wa > 0 for the fibre coordinates ya, we can define on the graded bundle F a globally defined weight vector field (Euler vector field) ∇F =

  • a

waya∂ya . The flow of the weight vector field extends to a smooth action R ∋ t → ht of multiplicative reals on F, ht(xµ, ya) = (xµ, twaya). Such an action h : R × F → F, ht ◦ hs = hts, we will call a homogeneity structure. A function f : F → R is called homogeneous of degree (weight) k if f (ht(x)) = tkf (x); similarly for the homogeneity of tensor fields. Morphisms of two homogeneity structures (Fi, hi), i = 1, 2, are defined as smooth maps Φ : F1 → F2 intertwining the R-actions: Φ ◦ h1

t = h2 t ◦ Φ. Consequently, a homogeneity substructure is a

smooth submanifold S invariant with respect to h, ht(S) ⊂ S.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 6 / 27

slide-46
SLIDE 46

Double Graded Bundles

The fundamental fact (c.f. Grabowski-Rotkiewicz) says that graded bundles and homogeneity structures are in fact equivalent concepts.

Theorem

For any homogeneity structure h on a manifold F, there is a smooth submanifold M = h0(F) ⊂ F, a non-negative integer k ∈ N, and an R-equivariant map Φk

h : F → TkF|M which identifies F with a graded

submanifold of the graded bundle TkF. In particular, there is an atlas on F consisting of local homogeneous functions. As two graded bundle structure on the same manifold are just two homogeneity structures, the obvious concept of compatibility leads to the following: A double graded bundle is a manifold equipped with two homogeneity structures h1, h2 which are compatible in the sense that h1

t ◦ h2 s = h2 s ◦ h1 t

for all s, t ∈ R . This covers of course the concept of a double vector bundle of Pradines and Mackenzie, and extends to n-tuple graded bundles in the obvious way.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 7 / 27

slide-47
SLIDE 47

Double Graded Bundles

The fundamental fact (c.f. Grabowski-Rotkiewicz) says that graded bundles and homogeneity structures are in fact equivalent concepts.

Theorem

For any homogeneity structure h on a manifold F, there is a smooth submanifold M = h0(F) ⊂ F, a non-negative integer k ∈ N, and an R-equivariant map Φk

h : F → TkF|M which identifies F with a graded

submanifold of the graded bundle TkF. In particular, there is an atlas on F consisting of local homogeneous functions. As two graded bundle structure on the same manifold are just two homogeneity structures, the obvious concept of compatibility leads to the following: A double graded bundle is a manifold equipped with two homogeneity structures h1, h2 which are compatible in the sense that h1

t ◦ h2 s = h2 s ◦ h1 t

for all s, t ∈ R . This covers of course the concept of a double vector bundle of Pradines and Mackenzie, and extends to n-tuple graded bundles in the obvious way.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 7 / 27

slide-48
SLIDE 48

Double Graded Bundles

The fundamental fact (c.f. Grabowski-Rotkiewicz) says that graded bundles and homogeneity structures are in fact equivalent concepts.

Theorem

For any homogeneity structure h on a manifold F, there is a smooth submanifold M = h0(F) ⊂ F, a non-negative integer k ∈ N, and an R-equivariant map Φk

h : F → TkF|M which identifies F with a graded

submanifold of the graded bundle TkF. In particular, there is an atlas on F consisting of local homogeneous functions. As two graded bundle structure on the same manifold are just two homogeneity structures, the obvious concept of compatibility leads to the following: A double graded bundle is a manifold equipped with two homogeneity structures h1, h2 which are compatible in the sense that h1

t ◦ h2 s = h2 s ◦ h1 t

for all s, t ∈ R . This covers of course the concept of a double vector bundle of Pradines and Mackenzie, and extends to n-tuple graded bundles in the obvious way.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 7 / 27

slide-49
SLIDE 49

Double Graded Bundles

The fundamental fact (c.f. Grabowski-Rotkiewicz) says that graded bundles and homogeneity structures are in fact equivalent concepts.

Theorem

For any homogeneity structure h on a manifold F, there is a smooth submanifold M = h0(F) ⊂ F, a non-negative integer k ∈ N, and an R-equivariant map Φk

h : F → TkF|M which identifies F with a graded

submanifold of the graded bundle TkF. In particular, there is an atlas on F consisting of local homogeneous functions. As two graded bundle structure on the same manifold are just two homogeneity structures, the obvious concept of compatibility leads to the following: A double graded bundle is a manifold equipped with two homogeneity structures h1, h2 which are compatible in the sense that h1

t ◦ h2 s = h2 s ◦ h1 t

for all s, t ∈ R . This covers of course the concept of a double vector bundle of Pradines and Mackenzie, and extends to n-tuple graded bundles in the obvious way.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 7 / 27

slide-50
SLIDE 50

Double Graded Bundles

The fundamental fact (c.f. Grabowski-Rotkiewicz) says that graded bundles and homogeneity structures are in fact equivalent concepts.

Theorem

For any homogeneity structure h on a manifold F, there is a smooth submanifold M = h0(F) ⊂ F, a non-negative integer k ∈ N, and an R-equivariant map Φk

h : F → TkF|M which identifies F with a graded

submanifold of the graded bundle TkF. In particular, there is an atlas on F consisting of local homogeneous functions. As two graded bundle structure on the same manifold are just two homogeneity structures, the obvious concept of compatibility leads to the following: A double graded bundle is a manifold equipped with two homogeneity structures h1, h2 which are compatible in the sense that h1

t ◦ h2 s = h2 s ◦ h1 t

for all s, t ∈ R . This covers of course the concept of a double vector bundle of Pradines and Mackenzie, and extends to n-tuple graded bundles in the obvious way.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 7 / 27

slide-51
SLIDE 51

Double Graded Bundles

The fundamental fact (c.f. Grabowski-Rotkiewicz) says that graded bundles and homogeneity structures are in fact equivalent concepts.

Theorem

For any homogeneity structure h on a manifold F, there is a smooth submanifold M = h0(F) ⊂ F, a non-negative integer k ∈ N, and an R-equivariant map Φk

h : F → TkF|M which identifies F with a graded

submanifold of the graded bundle TkF. In particular, there is an atlas on F consisting of local homogeneous functions. As two graded bundle structure on the same manifold are just two homogeneity structures, the obvious concept of compatibility leads to the following: A double graded bundle is a manifold equipped with two homogeneity structures h1, h2 which are compatible in the sense that h1

t ◦ h2 s = h2 s ◦ h1 t

for all s, t ∈ R . This covers of course the concept of a double vector bundle of Pradines and Mackenzie, and extends to n-tuple graded bundles in the obvious way.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 7 / 27

slide-52
SLIDE 52

Double graded bundles - examples

  • Lifts. If τ : F → M is a graded bundle of degree k, then TF and T∗F

carry canonical double graded bundle structure: one is the obvious vector bundle, the other is of degree k. A double graded bundle whose one structure is linear we will call a GL-bundle. There are also lifts of graded structures on F to TrF. In particular, if τ : E → M is a vector bundle, then TE and T∗E are double vector bundles. The latter is isomorphic with T∗E ∗. As a linear Poisson structure on E ∗ yields a map T∗E ∗ → TE ∗, a Lie algebroid structure on E can be encoded as a morphism of double vector bundles, ε : T∗E → TE ∗ (!) If τ : E → M is a vector bundle, then ∧kTE is canonically a GL-bundle: ∧kTE

♥♥♥♥

❘ ❘ ❘

E

◗ ◗ ◗ ◗

∧kTM

❧❧❧❧❧❧

M .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 8 / 27

slide-53
SLIDE 53

Double graded bundles - examples

  • Lifts. If τ : F → M is a graded bundle of degree k, then TF and T∗F

carry canonical double graded bundle structure: one is the obvious vector bundle, the other is of degree k. A double graded bundle whose one structure is linear we will call a GL-bundle. There are also lifts of graded structures on F to TrF. In particular, if τ : E → M is a vector bundle, then TE and T∗E are double vector bundles. The latter is isomorphic with T∗E ∗. As a linear Poisson structure on E ∗ yields a map T∗E ∗ → TE ∗, a Lie algebroid structure on E can be encoded as a morphism of double vector bundles, ε : T∗E → TE ∗ (!) If τ : E → M is a vector bundle, then ∧kTE is canonically a GL-bundle: ∧kTE

♥♥♥♥

❘ ❘ ❘

E

◗ ◗ ◗ ◗

∧kTM

❧❧❧❧❧❧

M .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 8 / 27

slide-54
SLIDE 54

Double graded bundles - examples

  • Lifts. If τ : F → M is a graded bundle of degree k, then TF and T∗F

carry canonical double graded bundle structure: one is the obvious vector bundle, the other is of degree k. A double graded bundle whose one structure is linear we will call a GL-bundle. There are also lifts of graded structures on F to TrF. In particular, if τ : E → M is a vector bundle, then TE and T∗E are double vector bundles. The latter is isomorphic with T∗E ∗. As a linear Poisson structure on E ∗ yields a map T∗E ∗ → TE ∗, a Lie algebroid structure on E can be encoded as a morphism of double vector bundles, ε : T∗E → TE ∗ (!) If τ : E → M is a vector bundle, then ∧kTE is canonically a GL-bundle: ∧kTE

♥♥♥♥

❘ ❘ ❘

E

◗ ◗ ◗ ◗

∧kTM

❧❧❧❧❧❧

M .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 8 / 27

slide-55
SLIDE 55

Double graded bundles - examples

  • Lifts. If τ : F → M is a graded bundle of degree k, then TF and T∗F

carry canonical double graded bundle structure: one is the obvious vector bundle, the other is of degree k. A double graded bundle whose one structure is linear we will call a GL-bundle. There are also lifts of graded structures on F to TrF. In particular, if τ : E → M is a vector bundle, then TE and T∗E are double vector bundles. The latter is isomorphic with T∗E ∗. As a linear Poisson structure on E ∗ yields a map T∗E ∗ → TE ∗, a Lie algebroid structure on E can be encoded as a morphism of double vector bundles, ε : T∗E → TE ∗ (!) If τ : E → M is a vector bundle, then ∧kTE is canonically a GL-bundle: ∧kTE

♥♥♥♥

❘ ❘ ❘

E

◗ ◗ ◗ ◗

∧kTM

❧❧❧❧❧❧

M .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 8 / 27

slide-56
SLIDE 56

Double graded bundles - examples

  • Lifts. If τ : F → M is a graded bundle of degree k, then TF and T∗F

carry canonical double graded bundle structure: one is the obvious vector bundle, the other is of degree k. A double graded bundle whose one structure is linear we will call a GL-bundle. There are also lifts of graded structures on F to TrF. In particular, if τ : E → M is a vector bundle, then TE and T∗E are double vector bundles. The latter is isomorphic with T∗E ∗. As a linear Poisson structure on E ∗ yields a map T∗E ∗ → TE ∗, a Lie algebroid structure on E can be encoded as a morphism of double vector bundles, ε : T∗E → TE ∗ (!) If τ : E → M is a vector bundle, then ∧kTE is canonically a GL-bundle: ∧kTE

♥♥♥♥

❘ ❘ ❘

E

◗ ◗ ◗ ◗

∧kTM

❧❧❧❧❧❧

M .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 8 / 27

slide-57
SLIDE 57

Double graded bundles - examples

  • Lifts. If τ : F → M is a graded bundle of degree k, then TF and T∗F

carry canonical double graded bundle structure: one is the obvious vector bundle, the other is of degree k. A double graded bundle whose one structure is linear we will call a GL-bundle. There are also lifts of graded structures on F to TrF. In particular, if τ : E → M is a vector bundle, then TE and T∗E are double vector bundles. The latter is isomorphic with T∗E ∗. As a linear Poisson structure on E ∗ yields a map T∗E ∗ → TE ∗, a Lie algebroid structure on E can be encoded as a morphism of double vector bundles, ε : T∗E → TE ∗ (!) If τ : E → M is a vector bundle, then ∧kTE is canonically a GL-bundle: ∧kTE

♥♥♥♥

❘ ❘ ❘

E

◗ ◗ ◗ ◗

∧kTM

❧❧❧❧❧❧

M .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 8 / 27

slide-58
SLIDE 58

Double graded bundles - examples

  • Lifts. If τ : F → M is a graded bundle of degree k, then TF and T∗F

carry canonical double graded bundle structure: one is the obvious vector bundle, the other is of degree k. A double graded bundle whose one structure is linear we will call a GL-bundle. There are also lifts of graded structures on F to TrF. In particular, if τ : E → M is a vector bundle, then TE and T∗E are double vector bundles. The latter is isomorphic with T∗E ∗. As a linear Poisson structure on E ∗ yields a map T∗E ∗ → TE ∗, a Lie algebroid structure on E can be encoded as a morphism of double vector bundles, ε : T∗E → TE ∗ (!) If τ : E → M is a vector bundle, then ∧kTE is canonically a GL-bundle: ∧kTE

♥♥♥♥

❘ ❘ ❘

E

◗ ◗ ◗ ◗

∧kTM

❧❧❧❧❧❧

M .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 8 / 27

slide-59
SLIDE 59

Double graded bundles - examples

  • Lifts. If τ : F → M is a graded bundle of degree k, then TF and T∗F

carry canonical double graded bundle structure: one is the obvious vector bundle, the other is of degree k. A double graded bundle whose one structure is linear we will call a GL-bundle. There are also lifts of graded structures on F to TrF. In particular, if τ : E → M is a vector bundle, then TE and T∗E are double vector bundles. The latter is isomorphic with T∗E ∗. As a linear Poisson structure on E ∗ yields a map T∗E ∗ → TE ∗, a Lie algebroid structure on E can be encoded as a morphism of double vector bundles, ε : T∗E → TE ∗ (!) If τ : E → M is a vector bundle, then ∧kTE is canonically a GL-bundle: ∧kTE

♥♥♥♥

❘ ❘ ❘

E

◗ ◗ ◗ ◗

∧kTM

❧❧❧❧❧❧

M .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 8 / 27

slide-60
SLIDE 60

Double graded bundles - examples

  • Lifts. If τ : F → M is a graded bundle of degree k, then TF and T∗F

carry canonical double graded bundle structure: one is the obvious vector bundle, the other is of degree k. A double graded bundle whose one structure is linear we will call a GL-bundle. There are also lifts of graded structures on F to TrF. In particular, if τ : E → M is a vector bundle, then TE and T∗E are double vector bundles. The latter is isomorphic with T∗E ∗. As a linear Poisson structure on E ∗ yields a map T∗E ∗ → TE ∗, a Lie algebroid structure on E can be encoded as a morphism of double vector bundles, ε : T∗E → TE ∗ (!) If τ : E → M is a vector bundle, then ∧kTE is canonically a GL-bundle: ∧kTE

♥♥♥♥

❘ ❘ ❘

E

◗ ◗ ◗ ◗

∧kTM

❧❧❧❧❧❧

M .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 8 / 27

slide-61
SLIDE 61

The Tulczyjew triple - Lagrangian side

M - positions,

TM - (kinematic)

configurations, L : TM → R - Lagrangian

T∗M - phase space

D = α−1

M (dL(TM))) = T L(TM) , image of the Tulczyjew differential T L ,

Legendre map: λL : TM → T∗M, λL(x, ˙ x) = (x, ∂L ∂ ˙ x ) , D =

  • (x, p, ˙

x, ˙ p) : p = ∂L ∂ ˙ x , ˙ p = ∂L ∂x

  • ,

whence the Euler-Lagrange equation:

∂L ∂x = d dt

∂L

∂ ˙ x

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 9 / 27

slide-62
SLIDE 62

The Tulczyjew triple - Lagrangian side

M - positions,

TM - (kinematic)

configurations, L : TM → R - Lagrangian

T∗M - phase space

D = α−1

M (dL(TM))) = T L(TM) , image of the Tulczyjew differential T L ,

Legendre map: λL : TM → T∗M, λL(x, ˙ x) = (x, ∂L ∂ ˙ x ) , D =

  • (x, p, ˙

x, ˙ p) : p = ∂L ∂ ˙ x , ˙ p = ∂L ∂x

  • ,

whence the Euler-Lagrange equation:

∂L ∂x = d dt

∂L

∂ ˙ x

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 9 / 27

slide-63
SLIDE 63

The Tulczyjew triple - Lagrangian side

M - positions,

TM - (kinematic)

configurations, L : TM → R - Lagrangian

T∗M - phase space

D = α−1

M (dL(TM))) = T L(TM) , image of the Tulczyjew differential T L ,

Legendre map: λL : TM → T∗M, λL(x, ˙ x) = (x, ∂L ∂ ˙ x ) , D =

  • (x, p, ˙

x, ˙ p) : p = ∂L ∂ ˙ x , ˙ p = ∂L ∂x

  • ,

whence the Euler-Lagrange equation:

∂L ∂x = d dt

∂L

∂ ˙ x

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 9 / 27

slide-64
SLIDE 64

The Tulczyjew triple - Lagrangian side

M - positions,

TM - (kinematic)

configurations, L : TM → R - Lagrangian

T∗M - phase space

D = α−1

M (dL(TM))) = T L(TM) , image of the Tulczyjew differential T L ,

Legendre map: λL : TM → T∗M, λL(x, ˙ x) = (x, ∂L ∂ ˙ x ) , D =

  • (x, p, ˙

x, ˙ p) : p = ∂L ∂ ˙ x , ˙ p = ∂L ∂x

  • ,

whence the Euler-Lagrange equation:

∂L ∂x = d dt

∂L

∂ ˙ x

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 9 / 27

slide-65
SLIDE 65

The Tulczyjew triple - Lagrangian side

M - positions,

TM - (kinematic)

configurations, L : TM → R - Lagrangian

T∗M - phase space

D = α−1

M (dL(TM))) = T L(TM) , image of the Tulczyjew differential T L ,

Legendre map: λL : TM → T∗M, λL(x, ˙ x) = (x, ∂L ∂ ˙ x ) , D =

  • (x, p, ˙

x, ˙ p) : p = ∂L ∂ ˙ x , ˙ p = ∂L ∂x

  • ,

whence the Euler-Lagrange equation:

∂L ∂x = d dt

∂L

∂ ˙ x

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 9 / 27

slide-66
SLIDE 66

The Tulczyjew triple - Lagrangian side

M - positions,

TM - (kinematic)

configurations, L : TM → R - Lagrangian

T∗M - phase space TT∗M

αM

❉ ❉ ❉ ❉ ☛☛☛☛☛☛☛☛☛☛ T∗TM

❉ ❉ ❉ ❉ ☛☛☛☛☛☛☛☛☛☛ TM

  • ☞☞☞☞☞☞☞☞☞☞

TM ☞☞☞☞☞☞☞☞☞☞ T∗M

❋ ❋ ❋ ❋ ❋ T∗M

❋ ❋ ❋ ❋ ❋

M

M

D = α−1

M (dL(TM))) = T L(TM) , image of the Tulczyjew differential T L ,

Legendre map: λL : TM → T∗M, λL(x, ˙ x) = (x, ∂L ∂ ˙ x ) , D =

  • (x, p, ˙

x, ˙ p) : p = ∂L ∂ ˙ x , ˙ p = ∂L ∂x

  • ,

whence the Euler-Lagrange equation:

∂L ∂x = d dt

∂L

∂ ˙ x

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 9 / 27

slide-67
SLIDE 67

The Tulczyjew triple - Lagrangian side

M - positions,

TM - (kinematic)

configurations, L : TM → R - Lagrangian

T∗M - phase space

D

TT∗M

αM

❉ ❉ ❉ ❉ ☛☛☛☛☛☛☛☛☛☛ T∗TM

πTM

❉ ❉ ❉ ❉ ☛☛☛☛☛☛☛☛☛☛ TM TM

dL

  • λL

❧❧❧❧❧❧❧❧❧❧

T L

❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳ T∗M T∗M

M M D = α−1

M (dL(TM))) = T L(TM) , image of the Tulczyjew differential T L ,

Legendre map: λL : TM → T∗M, λL(x, ˙ x) = (x, ∂L ∂ ˙ x ) , D =

  • (x, p, ˙

x, ˙ p) : p = ∂L ∂ ˙ x , ˙ p = ∂L ∂x

  • ,

whence the Euler-Lagrange equation:

∂L ∂x = d dt

∂L

∂ ˙ x

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 9 / 27

slide-68
SLIDE 68

The Tulczyjew triple - Lagrangian side

M - positions,

TM - (kinematic)

configurations, L : TM → R - Lagrangian

T∗M - phase space

D

TT∗M

αM

❉ ❉ ❉ ❉ ☛☛☛☛☛☛☛☛☛☛ T∗TM

πTM

❉ ❉ ❉ ❉ ☛☛☛☛☛☛☛☛☛☛ TM TM

dL

  • λL

❧❧❧❧❧❧❧❧❧❧

T L

❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳ T∗M T∗M

M M D = α−1

M (dL(TM))) = T L(TM) , image of the Tulczyjew differential T L ,

Legendre map: λL : TM → T∗M, λL(x, ˙ x) = (x, ∂L ∂ ˙ x ) , D =

  • (x, p, ˙

x, ˙ p) : p = ∂L ∂ ˙ x , ˙ p = ∂L ∂x

  • ,

whence the Euler-Lagrange equation:

∂L ∂x = d dt

∂L

∂ ˙ x

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 9 / 27

slide-69
SLIDE 69

The Tulczyjew triple - Lagrangian side

M - positions,

TM - (kinematic)

configurations, L : TM → R - Lagrangian

T∗M - phase space

D

TT∗M

αM

❉ ❉ ❉ ❉ ☛☛☛☛☛☛☛☛☛☛ T∗TM

πTM

❉ ❉ ❉ ❉ ☛☛☛☛☛☛☛☛☛☛ TM TM

dL

  • λL

❧❧❧❧❧❧❧❧❧❧

T L

❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳ T∗M T∗M

M M D = α−1

M (dL(TM))) = T L(TM) , image of the Tulczyjew differential T L ,

Legendre map: λL : TM → T∗M, λL(x, ˙ x) = (x, ∂L ∂ ˙ x ) , D =

  • (x, p, ˙

x, ˙ p) : p = ∂L ∂ ˙ x , ˙ p = ∂L ∂x

  • ,

whence the Euler-Lagrange equation:

∂L ∂x = d dt

∂L

∂ ˙ x

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 9 / 27

slide-70
SLIDE 70

The Tulczyjew triple - Lagrangian side

M - positions,

TM - (kinematic)

configurations, L : TM → R - Lagrangian

T∗M - phase space

D

TT∗M

αM

❉ ❉ ❉ ❉ ☛☛☛☛☛☛☛☛☛☛ T∗TM

πTM

❉ ❉ ❉ ❉ ☛☛☛☛☛☛☛☛☛☛ TM TM

dL

  • λL

❧❧❧❧❧❧❧❧❧❧

T L

❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳ T∗M T∗M

M M D = α−1

M (dL(TM))) = T L(TM) , image of the Tulczyjew differential T L ,

Legendre map: λL : TM → T∗M, λL(x, ˙ x) = (x, ∂L ∂ ˙ x ) , D =

  • (x, p, ˙

x, ˙ p) : p = ∂L ∂ ˙ x , ˙ p = ∂L ∂x

  • ,

whence the Euler-Lagrange equation:

∂L ∂x = d dt

∂L

∂ ˙ x

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 9 / 27

slide-71
SLIDE 71

The Tulczyjew triple - Hamiltonian side

H : T∗M → R D = β−1

M (dH(T∗M))

D =

  • (x, p, ˙

x, ˙ p) : ˙ p = −∂H ∂x , ˙ x = ∂H ∂p

  • ,

whence the Hamilton equations.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 10 / 27

slide-72
SLIDE 72

The Tulczyjew triple - Hamiltonian side

H : T∗M → R D = β−1

M (dH(T∗M))

D =

  • (x, p, ˙

x, ˙ p) : ˙ p = −∂H ∂x , ˙ x = ∂H ∂p

  • ,

whence the Hamilton equations.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 10 / 27

slide-73
SLIDE 73

The Tulczyjew triple - Hamiltonian side

H : T∗M → R

T∗T∗M

❋ ❋ ❋ ❋ ✠✠✠✠✠✠✠✠✠✠✠ TT∗M

❉ ❉ ❉ ❉ ☛☛☛☛☛☛☛☛☛☛

βM

  • TM

☛☛☛☛☛☛☛☛☛☛ TM ☞☞☞☞☞☞☞☞☞☞

  • T∗M

❍ ❍ ❍ ❍ ❍ T∗M

❋ ❋ ❋ ❋ ❋

  • M

M

  • D = β−1

M (dH(T∗M))

D =

  • (x, p, ˙

x, ˙ p) : ˙ p = −∂H ∂x , ˙ x = ∂H ∂p

  • ,

whence the Hamilton equations.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 10 / 27

slide-74
SLIDE 74

The Tulczyjew triple - Hamiltonian side

H : T∗M → R

T∗T∗M

❋ ❋ ❋ ❋ ✠✠✠✠✠✠✠✠✠✠✠ TT∗M

❉ ❉ ❉ ❉ ☛☛☛☛☛☛☛☛☛☛

βM

  • D
  • TM

TM T∗M

dH

  • T∗M

M M D = β−1

M (dH(T∗M))

D =

  • (x, p, ˙

x, ˙ p) : ˙ p = −∂H ∂x , ˙ x = ∂H ∂p

  • ,

whence the Hamilton equations.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 10 / 27

slide-75
SLIDE 75

The Tulczyjew triple - Hamiltonian side

H : T∗M → R

T∗T∗M

❋ ❋ ❋ ❋ ✠✠✠✠✠✠✠✠✠✠✠ TT∗M

❉ ❉ ❉ ❉ ☛☛☛☛☛☛☛☛☛☛

βM

  • D
  • TM

TM T∗M

dH

  • T∗M

M M D = β−1

M (dH(T∗M))

D =

  • (x, p, ˙

x, ˙ p) : ˙ p = −∂H ∂x , ˙ x = ∂H ∂p

  • ,

whence the Hamilton equations.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 10 / 27

slide-76
SLIDE 76

The Tulczyjew triple - Hamiltonian side

H : T∗M → R

T∗T∗M

❋ ❋ ❋ ❋ ✠✠✠✠✠✠✠✠✠✠✠ TT∗M

❉ ❉ ❉ ❉ ☛☛☛☛☛☛☛☛☛☛

βM

  • D
  • TM

TM T∗M

dH

  • T∗M

M M D = β−1

M (dH(T∗M))

D =

  • (x, p, ˙

x, ˙ p) : ˙ p = −∂H ∂x , ˙ x = ∂H ∂p

  • ,

whence the Hamilton equations.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 10 / 27

slide-77
SLIDE 77

Algebroid setting

DL

T∗E ∗

❊ ❊ ❊ ❊ ❊ ✡✡✡✡✡✡✡✡✡✡

˜ Π

TE ∗

  • ☛☛☛☛☛☛☛☛☛☛

T∗E

❈ ❈ ❈ ❈ ☛☛☛☛☛☛☛☛☛☛

ε

  • E

ρ

  • ☛☛☛☛☛☛☛☛☛☛

TM ✡✡✡✡✡✡✡✡✡✡

E

✌✌✌✌✌✌✌✌✌✌

ρ

  • E ∗
  • E ∗

❊ ❊ ❊ ❊

E ∗

❊ ❊ ❊ ❊

M

M

M

  • H : E ∗ −

→ R DH ⊂ T∗E ∗ D = T L(E) D = ˜ Π(dH(E ∗)) L : E − → R DL ⊂ T∗E

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 11 / 27

slide-78
SLIDE 78

Algebroid setting

DL

  • T∗E ∗

❊ ❊ ❊ ❊ ❊ ✡✡✡✡✡✡✡✡✡✡

˜ Π

TE ∗

  • ☛☛☛☛☛☛☛☛☛☛

T∗E

❈ ❈ ❈ ❈ ☛☛☛☛☛☛☛☛☛☛

ε

  • E

ρ

  • ☛☛☛☛☛☛☛☛☛☛

TM ✡✡✡✡✡✡✡✡✡✡

E

✌✌✌✌✌✌✌✌✌✌

ρ

  • dL
  • λL

❧❧❧❧❧❧❧❧❧❧❧

T L

❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳

E ∗

  • E ∗

❊ ❊ ❊ ❊

E ∗

❊ ❊ ❊ ❊

M

M

M

  • H : E ∗ −

→ R DH ⊂ T∗E ∗ D = T L(E) D = ˜ Π(dH(E ∗)) L : E − → R DL ⊂ T∗E

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 11 / 27

slide-79
SLIDE 79

Algebroid setting

D

  • DL
  • T∗E ∗

❊ ❊ ❊ ❊ ❊ ✡✡✡✡✡✡✡✡✡✡

˜ Π

TE ∗

  • ☛☛☛☛☛☛☛☛☛☛

T∗E

❈ ❈ ❈ ❈ ☛☛☛☛☛☛☛☛☛☛

ε

  • E

ρ

  • ☛☛☛☛☛☛☛☛☛☛

TM ✡✡✡✡✡✡✡✡✡✡

E

✌✌✌✌✌✌✌✌✌✌

ρ

  • dL
  • λL

❧❧❧❧❧❧❧❧❧❧❧

T L

❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳

E ∗

  • E ∗

❊ ❊ ❊ ❊

E ∗

❊ ❊ ❊ ❊

M

M

M

  • H : E ∗ −

→ R DH ⊂ T∗E ∗ D = T L(E) D = ˜ Π(dH(E ∗)) L : E − → R DL ⊂ T∗E

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 11 / 27

slide-80
SLIDE 80

Algebroid setting

DH

  • D
  • DL
  • T∗E ∗

❊ ❊ ❊ ❊ ❊ ✡✡✡✡✡✡✡✡✡✡

˜ Π

TE ∗

  • ☛☛☛☛☛☛☛☛☛☛

T∗E

❈ ❈ ❈ ❈ ☛☛☛☛☛☛☛☛☛☛

ε

  • E

ρ

  • ☛☛☛☛☛☛☛☛☛☛

TM ✡✡✡✡✡✡✡✡✡✡

E

✌✌✌✌✌✌✌✌✌✌

ρ

  • dL
  • λL

❧❧❧❧❧❧❧❧❧❧❧

T L

❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳❳

E ∗

  • dH
  • E ∗

❊ ❊ ❊ ❊

E ∗

❊ ❊ ❊ ❊

M

M

M

  • H : E ∗ −

→ R DH ⊂ T∗E ∗ D = T L(E) D = ˜ Π(dH(E ∗)) L : E − → R DL ⊂ T∗E

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 11 / 27

slide-81
SLIDE 81

Algebroid setting with vakonomic constraints

D

  • SL
  • TE ∗

❍ ❍ ❍ ❍ ✡✡✡✡✡✡✡✡✡✡✡ T∗E

❑ ❑ ❑ ❑ ❑ ✡✡✡✡✡✡✡✡✡✡✡

ε

  • TM

✠✠✠✠✠✠✠✠✠✠✠

E ⊃ S

✆✆✆✆✆✆✆✆✆✆✆

ρ

  • SL
  • λL

❥❥❥❥❥❥❥❥❥❥❥❥

E ∗

❋ ❋ ❋ ❋ ❋

E ∗

❋ ❋ ❋ ❋ ❋

M M

  • where SL is the lagrangian submanifold in T∗E induced by the Lagrangian
  • n the constraint S, and SL : S → T∗E is the corresponding relation,

SL = {αe ∈ T∗

eE : e ∈ S and αe, ve = dL(ve) for every ve ∈ TeS} .

The vakonomically constrained phase dynamics is just D = ε(SL) ⊂ TE ∗.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 12 / 27

slide-82
SLIDE 82

Algebroid setting with vakonomic constraints

D

  • SL
  • TE ∗

❍ ❍ ❍ ❍ ✡✡✡✡✡✡✡✡✡✡✡ T∗E

❑ ❑ ❑ ❑ ❑ ✡✡✡✡✡✡✡✡✡✡✡

ε

  • TM

✠✠✠✠✠✠✠✠✠✠✠

E ⊃ S

✆✆✆✆✆✆✆✆✆✆✆

ρ

  • SL
  • λL

❥❥❥❥❥❥❥❥❥❥❥❥

E ∗

❋ ❋ ❋ ❋ ❋

E ∗

❋ ❋ ❋ ❋ ❋

M M

  • where SL is the lagrangian submanifold in T∗E induced by the Lagrangian
  • n the constraint S, and SL : S → T∗E is the corresponding relation,

SL = {αe ∈ T∗

eE : e ∈ S and αe, ve = dL(ve) for every ve ∈ TeS} .

The vakonomically constrained phase dynamics is just D = ε(SL) ⊂ TE ∗.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 12 / 27

slide-83
SLIDE 83

Algebroid setting with vakonomic constraints

D

  • SL
  • TE ∗

❍ ❍ ❍ ❍ ✡✡✡✡✡✡✡✡✡✡✡ T∗E

❑ ❑ ❑ ❑ ❑ ✡✡✡✡✡✡✡✡✡✡✡

ε

  • TM

✠✠✠✠✠✠✠✠✠✠✠

E ⊃ S

✆✆✆✆✆✆✆✆✆✆✆

ρ

  • SL
  • λL

❥❥❥❥❥❥❥❥❥❥❥❥

E ∗

❋ ❋ ❋ ❋ ❋

E ∗

❋ ❋ ❋ ❋ ❋

M M

  • where SL is the lagrangian submanifold in T∗E induced by the Lagrangian
  • n the constraint S, and SL : S → T∗E is the corresponding relation,

SL = {αe ∈ T∗

eE : e ∈ S and αe, ve = dL(ve) for every ve ∈ TeS} .

The vakonomically constrained phase dynamics is just D = ε(SL) ⊂ TE ∗.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 12 / 27

slide-84
SLIDE 84

Algebroid setting with vakonomic constraints

D

  • SL
  • TE ∗

❍ ❍ ❍ ❍ ✡✡✡✡✡✡✡✡✡✡✡ T∗E

❑ ❑ ❑ ❑ ❑ ✡✡✡✡✡✡✡✡✡✡✡

ε

  • TM

✠✠✠✠✠✠✠✠✠✠✠

E ⊃ S

✆✆✆✆✆✆✆✆✆✆✆

ρ

  • SL
  • λL

❥❥❥❥❥❥❥❥❥❥❥❥

E ∗

❋ ❋ ❋ ❋ ❋

E ∗

❋ ❋ ❋ ❋ ❋

M M

  • where SL is the lagrangian submanifold in T∗E induced by the Lagrangian
  • n the constraint S, and SL : S → T∗E is the corresponding relation,

SL = {αe ∈ T∗

eE : e ∈ S and αe, ve = dL(ve) for every ve ∈ TeS} .

The vakonomically constrained phase dynamics is just D = ε(SL) ⊂ TE ∗.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 12 / 27

slide-85
SLIDE 85

Higher order Lagrangians

The mechanics with a higher order Lagrangian L : TkQ → R is traditionally constructed as a vakonomic mechanics, thanks to the canonical embedding of of the higher tangent bundle TkQ into the tangent bundle TTk−1Q as an affine subbundle of holonomic vectors. Thus we work with the standard Tulczyjew triple for TM, where M = Tk−1Q, with the presence of vakonomic constraint TkQ ⊂ TTk−1Q:

TT∗Tk−1Q rrrr

✾ ✾ ✾ ✾ ✾ ✾ ✾ ✾ ✾ T∗TTk−1Q

  • T∗TkQ

  • ✐✐✐✐✐✐✐✐
  • T∗Tk−1Q

✿ ✿ ✿ ✿ ✿ ✿ ✿ ✿ ✿ Tk−1Q ×Q T∗Q

✻ ✻ ✻ ✻ ✻ ✻ ✻ ✻ TTk−1Q s s s s TkQ ②②②

  • Tk−1Q

Tk−1Q

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 13 / 27

slide-86
SLIDE 86

Higher order Lagrangians

The mechanics with a higher order Lagrangian L : TkQ → R is traditionally constructed as a vakonomic mechanics, thanks to the canonical embedding of of the higher tangent bundle TkQ into the tangent bundle TTk−1Q as an affine subbundle of holonomic vectors. Thus we work with the standard Tulczyjew triple for TM, where M = Tk−1Q, with the presence of vakonomic constraint TkQ ⊂ TTk−1Q:

TT∗Tk−1Q rrrr

✾ ✾ ✾ ✾ ✾ ✾ ✾ ✾ ✾ T∗TTk−1Q

  • T∗TkQ

  • ✐✐✐✐✐✐✐✐
  • T∗Tk−1Q

✿ ✿ ✿ ✿ ✿ ✿ ✿ ✿ ✿ Tk−1Q ×Q T∗Q

✻ ✻ ✻ ✻ ✻ ✻ ✻ ✻ TTk−1Q s s s s TkQ ②②②

  • Tk−1Q

Tk−1Q

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 13 / 27

slide-87
SLIDE 87

Higher order Lagrangians

The mechanics with a higher order Lagrangian L : TkQ → R is traditionally constructed as a vakonomic mechanics, thanks to the canonical embedding of of the higher tangent bundle TkQ into the tangent bundle TTk−1Q as an affine subbundle of holonomic vectors. Thus we work with the standard Tulczyjew triple for TM, where M = Tk−1Q, with the presence of vakonomic constraint TkQ ⊂ TTk−1Q:

TT∗Tk−1Q rrrr

✾ ✾ ✾ ✾ ✾ ✾ ✾ ✾ ✾ T∗TTk−1Q

  • T∗TkQ

  • ✐✐✐✐✐✐✐✐
  • T∗Tk−1Q

✿ ✿ ✿ ✿ ✿ ✿ ✿ ✿ ✿ Tk−1Q ×Q T∗Q

✻ ✻ ✻ ✻ ✻ ✻ ✻ ✻ TTk−1Q s s s s TkQ ②②②

  • Tk−1Q

Tk−1Q

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 13 / 27

slide-88
SLIDE 88

Higher order Lagrangians

The mechanics with a higher order Lagrangian L : TkQ → R is traditionally constructed as a vakonomic mechanics, thanks to the canonical embedding of of the higher tangent bundle TkQ into the tangent bundle TTk−1Q as an affine subbundle of holonomic vectors. Thus we work with the standard Tulczyjew triple for TM, where M = Tk−1Q, with the presence of vakonomic constraint TkQ ⊂ TTk−1Q:

TT∗Tk−1Q rrrr

✾ ✾ ✾ ✾ ✾ ✾ ✾ ✾ ✾ T∗TTk−1Q

  • T∗TkQ

  • ✐✐✐✐✐✐✐✐
  • T∗Tk−1Q

✿ ✿ ✿ ✿ ✿ ✿ ✿ ✿ ✿ Tk−1Q ×Q T∗Q

✻ ✻ ✻ ✻ ✻ ✻ ✻ ✻ TTk−1Q s s s s TkQ ②②②

  • Tk−1Q

Tk−1Q

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 13 / 27

slide-89
SLIDE 89

Higher order Lagrangians

The mechanics with a higher order Lagrangian L : TkQ → R is traditionally constructed as a vakonomic mechanics, thanks to the canonical embedding of of the higher tangent bundle TkQ into the tangent bundle TTk−1Q as an affine subbundle of holonomic vectors. Thus we work with the standard Tulczyjew triple for TM, where M = Tk−1Q, with the presence of vakonomic constraint TkQ ⊂ TTk−1Q:

TT∗Tk−1Q rrrr

✾ ✾ ✾ ✾ ✾ ✾ ✾ ✾ ✾ T∗TTk−1Q

  • T∗TkQ

  • ✐✐✐✐✐✐✐✐
  • T∗Tk−1Q

✿ ✿ ✿ ✿ ✿ ✿ ✿ ✿ ✿ Tk−1Q ×Q T∗Q

✻ ✻ ✻ ✻ ✻ ✻ ✻ ✻ TTk−1Q s s s s TkQ ②②②

  • Tk−1Q

Tk−1Q

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 13 / 27

slide-90
SLIDE 90

Higher order Euler-Lagrange equations

The Lagrangian function L = L(q, . . . ,

(k)

q ) generates the phase dynamics D =    (v, p, ˙ v, ˙ p) : ˙ vi−1 = vi, ˙ pi + pi−1 = ∂L ∂

(i)

q , ˙ p0 = ∂L ∂q , pk−1 = ∂L ∂

(k)

q    . This leads to the higher Euler-Lagrange equations in the traditional form:

(i)

q = diq dti , i = 1, . . . , k , 0 = ∂L ∂q − d dt ∂L ∂ ˙ q

  • + · · · + (−1)k dk

dtk   ∂L ∂

(k)

q   . These equations can be viewed as a system of differential equations of

  • rder k on TkQ or, which is the standard point of view, as ordinary

differential equation of order 2k on Q.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 14 / 27

slide-91
SLIDE 91

Higher order Euler-Lagrange equations

The Lagrangian function L = L(q, . . . ,

(k)

q ) generates the phase dynamics D =    (v, p, ˙ v, ˙ p) : ˙ vi−1 = vi, ˙ pi + pi−1 = ∂L ∂

(i)

q , ˙ p0 = ∂L ∂q , pk−1 = ∂L ∂

(k)

q    . This leads to the higher Euler-Lagrange equations in the traditional form:

(i)

q = diq dti , i = 1, . . . , k , 0 = ∂L ∂q − d dt ∂L ∂ ˙ q

  • + · · · + (−1)k dk

dtk   ∂L ∂

(k)

q   . These equations can be viewed as a system of differential equations of

  • rder k on TkQ or, which is the standard point of view, as ordinary

differential equation of order 2k on Q.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 14 / 27

slide-92
SLIDE 92

Higher order Euler-Lagrange equations

The Lagrangian function L = L(q, . . . ,

(k)

q ) generates the phase dynamics D =    (v, p, ˙ v, ˙ p) : ˙ vi−1 = vi, ˙ pi + pi−1 = ∂L ∂

(i)

q , ˙ p0 = ∂L ∂q , pk−1 = ∂L ∂

(k)

q    . This leads to the higher Euler-Lagrange equations in the traditional form:

(i)

q = diq dti , i = 1, . . . , k , 0 = ∂L ∂q − d dt ∂L ∂ ˙ q

  • + · · · + (−1)k dk

dtk   ∂L ∂

(k)

q   . These equations can be viewed as a system of differential equations of

  • rder k on TkQ or, which is the standard point of view, as ordinary

differential equation of order 2k on Q.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 14 / 27

slide-93
SLIDE 93

Higher order Euler-Lagrange equations

The Lagrangian function L = L(q, . . . ,

(k)

q ) generates the phase dynamics D =    (v, p, ˙ v, ˙ p) : ˙ vi−1 = vi, ˙ pi + pi−1 = ∂L ∂

(i)

q , ˙ p0 = ∂L ∂q , pk−1 = ∂L ∂

(k)

q    . This leads to the higher Euler-Lagrange equations in the traditional form:

(i)

q = diq dti , i = 1, . . . , k , 0 = ∂L ∂q − d dt ∂L ∂ ˙ q

  • + · · · + (−1)k dk

dtk   ∂L ∂

(k)

q   . These equations can be viewed as a system of differential equations of

  • rder k on TkQ or, which is the standard point of view, as ordinary

differential equation of order 2k on Q.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 14 / 27

slide-94
SLIDE 94

Higher order Euler-Lagrange equations

The Lagrangian function L = L(q, . . . ,

(k)

q ) generates the phase dynamics D =    (v, p, ˙ v, ˙ p) : ˙ vi−1 = vi, ˙ pi + pi−1 = ∂L ∂

(i)

q , ˙ p0 = ∂L ∂q , pk−1 = ∂L ∂

(k)

q    . This leads to the higher Euler-Lagrange equations in the traditional form:

(i)

q = diq dti , i = 1, . . . , k , 0 = ∂L ∂q − d dt ∂L ∂ ˙ q

  • + · · · + (−1)k dk

dtk   ∂L ∂

(k)

q   . These equations can be viewed as a system of differential equations of

  • rder k on TkQ or, which is the standard point of view, as ordinary

differential equation of order 2k on Q.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 14 / 27

slide-95
SLIDE 95

Higher order Euler-Lagrange equations

The Lagrangian function L = L(q, . . . ,

(k)

q ) generates the phase dynamics D =    (v, p, ˙ v, ˙ p) : ˙ vi−1 = vi, ˙ pi + pi−1 = ∂L ∂

(i)

q , ˙ p0 = ∂L ∂q , pk−1 = ∂L ∂

(k)

q    . This leads to the higher Euler-Lagrange equations in the traditional form:

(i)

q = diq dti , i = 1, . . . , k , 0 = ∂L ∂q − d dt ∂L ∂ ˙ q

  • + · · · + (−1)k dk

dtk   ∂L ∂

(k)

q   . These equations can be viewed as a system of differential equations of

  • rder k on TkQ or, which is the standard point of view, as ordinary

differential equation of order 2k on Q.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 14 / 27

slide-96
SLIDE 96

Higher order Euler-Lagrange equations

The Lagrangian function L = L(q, . . . ,

(k)

q ) generates the phase dynamics D =    (v, p, ˙ v, ˙ p) : ˙ vi−1 = vi, ˙ pi + pi−1 = ∂L ∂

(i)

q , ˙ p0 = ∂L ∂q , pk−1 = ∂L ∂

(k)

q    . This leads to the higher Euler-Lagrange equations in the traditional form:

(i)

q = diq dti , i = 1, . . . , k , 0 = ∂L ∂q − d dt ∂L ∂ ˙ q

  • + · · · + (−1)k dk

dtk   ∂L ∂

(k)

q   . These equations can be viewed as a system of differential equations of

  • rder k on TkQ or, which is the standard point of view, as ordinary

differential equation of order 2k on Q.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 14 / 27

slide-97
SLIDE 97

Higher order Euler-Lagrange equations

The Lagrangian function L = L(q, . . . ,

(k)

q ) generates the phase dynamics D =    (v, p, ˙ v, ˙ p) : ˙ vi−1 = vi, ˙ pi + pi−1 = ∂L ∂

(i)

q , ˙ p0 = ∂L ∂q , pk−1 = ∂L ∂

(k)

q    . This leads to the higher Euler-Lagrange equations in the traditional form:

(i)

q = diq dti , i = 1, . . . , k , 0 = ∂L ∂q − d dt ∂L ∂ ˙ q

  • + · · · + (−1)k dk

dtk   ∂L ∂

(k)

q   . These equations can be viewed as a system of differential equations of

  • rder k on TkQ or, which is the standard point of view, as ordinary

differential equation of order 2k on Q.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 14 / 27

slide-98
SLIDE 98

Linearisation of graded bundles

The possibility of constructing mechanics on graded bundles is based on the following generalization of the embedding TkQ ֒ → TTk−1Q.

Theorem (Bruce-Grabowska-Grabowski)

There is a canonical functor from the category of graded bundles into the category of GL-bundles which assigns, for an arbitrary graded bundle Fk of degree k, a canonical GL-bundle D(Fk) which is linear over Fk−1, called the linearisation of Fk, together with a graded embedding ι : Fk ֒ → D(Fk)

  • f Fk as an affine subbundle of the vector bundle D(Fk).

Elements of Fk ⊂ D(Fk) may be viewed as ‘holonomic vectors’ in the linear-graded bundle D(Fk). Another geometric part we need is a (Lie) algebroid structure on the vector bundle D(Fk) → Fk−1, compatible with the second graded structure (homogeneity). We will call such GL-bundles D weighted (Lie) algebroids and view them as abstract generalizations of the Lie algebroid TTk−1M. Such D is called a VB-algebroid if it is a double vector bundle.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 15 / 27

slide-99
SLIDE 99

Linearisation of graded bundles

The possibility of constructing mechanics on graded bundles is based on the following generalization of the embedding TkQ ֒ → TTk−1Q.

Theorem (Bruce-Grabowska-Grabowski)

There is a canonical functor from the category of graded bundles into the category of GL-bundles which assigns, for an arbitrary graded bundle Fk of degree k, a canonical GL-bundle D(Fk) which is linear over Fk−1, called the linearisation of Fk, together with a graded embedding ι : Fk ֒ → D(Fk)

  • f Fk as an affine subbundle of the vector bundle D(Fk).

Elements of Fk ⊂ D(Fk) may be viewed as ‘holonomic vectors’ in the linear-graded bundle D(Fk). Another geometric part we need is a (Lie) algebroid structure on the vector bundle D(Fk) → Fk−1, compatible with the second graded structure (homogeneity). We will call such GL-bundles D weighted (Lie) algebroids and view them as abstract generalizations of the Lie algebroid TTk−1M. Such D is called a VB-algebroid if it is a double vector bundle.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 15 / 27

slide-100
SLIDE 100

Linearisation of graded bundles

The possibility of constructing mechanics on graded bundles is based on the following generalization of the embedding TkQ ֒ → TTk−1Q.

Theorem (Bruce-Grabowska-Grabowski)

There is a canonical functor from the category of graded bundles into the category of GL-bundles which assigns, for an arbitrary graded bundle Fk of degree k, a canonical GL-bundle D(Fk) which is linear over Fk−1, called the linearisation of Fk, together with a graded embedding ι : Fk ֒ → D(Fk)

  • f Fk as an affine subbundle of the vector bundle D(Fk).

Elements of Fk ⊂ D(Fk) may be viewed as ‘holonomic vectors’ in the linear-graded bundle D(Fk). Another geometric part we need is a (Lie) algebroid structure on the vector bundle D(Fk) → Fk−1, compatible with the second graded structure (homogeneity). We will call such GL-bundles D weighted (Lie) algebroids and view them as abstract generalizations of the Lie algebroid TTk−1M. Such D is called a VB-algebroid if it is a double vector bundle.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 15 / 27

slide-101
SLIDE 101

Linearisation of graded bundles

The possibility of constructing mechanics on graded bundles is based on the following generalization of the embedding TkQ ֒ → TTk−1Q.

Theorem (Bruce-Grabowska-Grabowski)

There is a canonical functor from the category of graded bundles into the category of GL-bundles which assigns, for an arbitrary graded bundle Fk of degree k, a canonical GL-bundle D(Fk) which is linear over Fk−1, called the linearisation of Fk, together with a graded embedding ι : Fk ֒ → D(Fk)

  • f Fk as an affine subbundle of the vector bundle D(Fk).

Elements of Fk ⊂ D(Fk) may be viewed as ‘holonomic vectors’ in the linear-graded bundle D(Fk). Another geometric part we need is a (Lie) algebroid structure on the vector bundle D(Fk) → Fk−1, compatible with the second graded structure (homogeneity). We will call such GL-bundles D weighted (Lie) algebroids and view them as abstract generalizations of the Lie algebroid TTk−1M. Such D is called a VB-algebroid if it is a double vector bundle.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 15 / 27

slide-102
SLIDE 102

Linearisation of graded bundles

The possibility of constructing mechanics on graded bundles is based on the following generalization of the embedding TkQ ֒ → TTk−1Q.

Theorem (Bruce-Grabowska-Grabowski)

There is a canonical functor from the category of graded bundles into the category of GL-bundles which assigns, for an arbitrary graded bundle Fk of degree k, a canonical GL-bundle D(Fk) which is linear over Fk−1, called the linearisation of Fk, together with a graded embedding ι : Fk ֒ → D(Fk)

  • f Fk as an affine subbundle of the vector bundle D(Fk).

Elements of Fk ⊂ D(Fk) may be viewed as ‘holonomic vectors’ in the linear-graded bundle D(Fk). Another geometric part we need is a (Lie) algebroid structure on the vector bundle D(Fk) → Fk−1, compatible with the second graded structure (homogeneity). We will call such GL-bundles D weighted (Lie) algebroids and view them as abstract generalizations of the Lie algebroid TTk−1M. Such D is called a VB-algebroid if it is a double vector bundle.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 15 / 27

slide-103
SLIDE 103

Linearisation of graded bundles

The possibility of constructing mechanics on graded bundles is based on the following generalization of the embedding TkQ ֒ → TTk−1Q.

Theorem (Bruce-Grabowska-Grabowski)

There is a canonical functor from the category of graded bundles into the category of GL-bundles which assigns, for an arbitrary graded bundle Fk of degree k, a canonical GL-bundle D(Fk) which is linear over Fk−1, called the linearisation of Fk, together with a graded embedding ι : Fk ֒ → D(Fk)

  • f Fk as an affine subbundle of the vector bundle D(Fk).

Elements of Fk ⊂ D(Fk) may be viewed as ‘holonomic vectors’ in the linear-graded bundle D(Fk). Another geometric part we need is a (Lie) algebroid structure on the vector bundle D(Fk) → Fk−1, compatible with the second graded structure (homogeneity). We will call such GL-bundles D weighted (Lie) algebroids and view them as abstract generalizations of the Lie algebroid TTk−1M. Such D is called a VB-algebroid if it is a double vector bundle.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 15 / 27

slide-104
SLIDE 104

Linearisation of graded bundles

The possibility of constructing mechanics on graded bundles is based on the following generalization of the embedding TkQ ֒ → TTk−1Q.

Theorem (Bruce-Grabowska-Grabowski)

There is a canonical functor from the category of graded bundles into the category of GL-bundles which assigns, for an arbitrary graded bundle Fk of degree k, a canonical GL-bundle D(Fk) which is linear over Fk−1, called the linearisation of Fk, together with a graded embedding ι : Fk ֒ → D(Fk)

  • f Fk as an affine subbundle of the vector bundle D(Fk).

Elements of Fk ⊂ D(Fk) may be viewed as ‘holonomic vectors’ in the linear-graded bundle D(Fk). Another geometric part we need is a (Lie) algebroid structure on the vector bundle D(Fk) → Fk−1, compatible with the second graded structure (homogeneity). We will call such GL-bundles D weighted (Lie) algebroids and view them as abstract generalizations of the Lie algebroid TTk−1M. Such D is called a VB-algebroid if it is a double vector bundle.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 15 / 27

slide-105
SLIDE 105

Linearisation of graded bundles

The possibility of constructing mechanics on graded bundles is based on the following generalization of the embedding TkQ ֒ → TTk−1Q.

Theorem (Bruce-Grabowska-Grabowski)

There is a canonical functor from the category of graded bundles into the category of GL-bundles which assigns, for an arbitrary graded bundle Fk of degree k, a canonical GL-bundle D(Fk) which is linear over Fk−1, called the linearisation of Fk, together with a graded embedding ι : Fk ֒ → D(Fk)

  • f Fk as an affine subbundle of the vector bundle D(Fk).

Elements of Fk ⊂ D(Fk) may be viewed as ‘holonomic vectors’ in the linear-graded bundle D(Fk). Another geometric part we need is a (Lie) algebroid structure on the vector bundle D(Fk) → Fk−1, compatible with the second graded structure (homogeneity). We will call such GL-bundles D weighted (Lie) algebroids and view them as abstract generalizations of the Lie algebroid TTk−1M. Such D is called a VB-algebroid if it is a double vector bundle.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 15 / 27

slide-106
SLIDE 106

Linearisation of graded bundles

The possibility of constructing mechanics on graded bundles is based on the following generalization of the embedding TkQ ֒ → TTk−1Q.

Theorem (Bruce-Grabowska-Grabowski)

There is a canonical functor from the category of graded bundles into the category of GL-bundles which assigns, for an arbitrary graded bundle Fk of degree k, a canonical GL-bundle D(Fk) which is linear over Fk−1, called the linearisation of Fk, together with a graded embedding ι : Fk ֒ → D(Fk)

  • f Fk as an affine subbundle of the vector bundle D(Fk).

Elements of Fk ⊂ D(Fk) may be viewed as ‘holonomic vectors’ in the linear-graded bundle D(Fk). Another geometric part we need is a (Lie) algebroid structure on the vector bundle D(Fk) → Fk−1, compatible with the second graded structure (homogeneity). We will call such GL-bundles D weighted (Lie) algebroids and view them as abstract generalizations of the Lie algebroid TTk−1M. Such D is called a VB-algebroid if it is a double vector bundle.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 15 / 27

slide-107
SLIDE 107

Linearisation of graded bundles

The possibility of constructing mechanics on graded bundles is based on the following generalization of the embedding TkQ ֒ → TTk−1Q.

Theorem (Bruce-Grabowska-Grabowski)

There is a canonical functor from the category of graded bundles into the category of GL-bundles which assigns, for an arbitrary graded bundle Fk of degree k, a canonical GL-bundle D(Fk) which is linear over Fk−1, called the linearisation of Fk, together with a graded embedding ι : Fk ֒ → D(Fk)

  • f Fk as an affine subbundle of the vector bundle D(Fk).

Elements of Fk ⊂ D(Fk) may be viewed as ‘holonomic vectors’ in the linear-graded bundle D(Fk). Another geometric part we need is a (Lie) algebroid structure on the vector bundle D(Fk) → Fk−1, compatible with the second graded structure (homogeneity). We will call such GL-bundles D weighted (Lie) algebroids and view them as abstract generalizations of the Lie algebroid TTk−1M. Such D is called a VB-algebroid if it is a double vector bundle.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 15 / 27

slide-108
SLIDE 108

Weighted Lie algebroids out of reductions

Let G ⇒ M be a Lie groupoid and consider the subbundle TkGs ⊂ TkG consisting of all higher order velocities tangent to source-leaves. The bundle Fk = Ak(G) := TkGs

  • M ,

inherits graded bundle structure of degree k as a graded subbundle of

  • TkG. Of course, A = A1(G) can be identified with the Lie algebroid of G.

Theorem

The linearisation of Ak(G) is given as D(Ak(G)) ≃ {(Y , Z) ∈ A(G) ×M TAk−1(G)| ρ(Y ) = Tτ(Z)} , viewed as a vector bundle over Ak−1(G) with respect to the obvious projection of part Z onto Ak−1(G), where ρ : A(G) → TM is the standard anchor of the Lie algebroid and τ : Ak−1(G) → M is the obvious projection. Moreover, the above bundle is canonically a weighted Lie algebroid, a Lie algebroid prolongation in the sense of Popescu and Mart´ ınez.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 16 / 27

slide-109
SLIDE 109

Weighted Lie algebroids out of reductions

Let G ⇒ M be a Lie groupoid and consider the subbundle TkGs ⊂ TkG consisting of all higher order velocities tangent to source-leaves. The bundle Fk = Ak(G) := TkGs

  • M ,

inherits graded bundle structure of degree k as a graded subbundle of

  • TkG. Of course, A = A1(G) can be identified with the Lie algebroid of G.

Theorem

The linearisation of Ak(G) is given as D(Ak(G)) ≃ {(Y , Z) ∈ A(G) ×M TAk−1(G)| ρ(Y ) = Tτ(Z)} , viewed as a vector bundle over Ak−1(G) with respect to the obvious projection of part Z onto Ak−1(G), where ρ : A(G) → TM is the standard anchor of the Lie algebroid and τ : Ak−1(G) → M is the obvious projection. Moreover, the above bundle is canonically a weighted Lie algebroid, a Lie algebroid prolongation in the sense of Popescu and Mart´ ınez.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 16 / 27

slide-110
SLIDE 110

Weighted Lie algebroids out of reductions

Let G ⇒ M be a Lie groupoid and consider the subbundle TkGs ⊂ TkG consisting of all higher order velocities tangent to source-leaves. The bundle Fk = Ak(G) := TkGs

  • M ,

inherits graded bundle structure of degree k as a graded subbundle of

  • TkG. Of course, A = A1(G) can be identified with the Lie algebroid of G.

Theorem

The linearisation of Ak(G) is given as D(Ak(G)) ≃ {(Y , Z) ∈ A(G) ×M TAk−1(G)| ρ(Y ) = Tτ(Z)} , viewed as a vector bundle over Ak−1(G) with respect to the obvious projection of part Z onto Ak−1(G), where ρ : A(G) → TM is the standard anchor of the Lie algebroid and τ : Ak−1(G) → M is the obvious projection. Moreover, the above bundle is canonically a weighted Lie algebroid, a Lie algebroid prolongation in the sense of Popescu and Mart´ ınez.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 16 / 27

slide-111
SLIDE 111

Weighted Lie algebroids out of reductions

Let G ⇒ M be a Lie groupoid and consider the subbundle TkGs ⊂ TkG consisting of all higher order velocities tangent to source-leaves. The bundle Fk = Ak(G) := TkGs

  • M ,

inherits graded bundle structure of degree k as a graded subbundle of

  • TkG. Of course, A = A1(G) can be identified with the Lie algebroid of G.

Theorem

The linearisation of Ak(G) is given as D(Ak(G)) ≃ {(Y , Z) ∈ A(G) ×M TAk−1(G)| ρ(Y ) = Tτ(Z)} , viewed as a vector bundle over Ak−1(G) with respect to the obvious projection of part Z onto Ak−1(G), where ρ : A(G) → TM is the standard anchor of the Lie algebroid and τ : Ak−1(G) → M is the obvious projection. Moreover, the above bundle is canonically a weighted Lie algebroid, a Lie algebroid prolongation in the sense of Popescu and Mart´ ınez.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 16 / 27

slide-112
SLIDE 112

Weighted Lie algebroids out of reductions

Let G ⇒ M be a Lie groupoid and consider the subbundle TkGs ⊂ TkG consisting of all higher order velocities tangent to source-leaves. The bundle Fk = Ak(G) := TkGs

  • M ,

inherits graded bundle structure of degree k as a graded subbundle of

  • TkG. Of course, A = A1(G) can be identified with the Lie algebroid of G.

Theorem

The linearisation of Ak(G) is given as D(Ak(G)) ≃ {(Y , Z) ∈ A(G) ×M TAk−1(G)| ρ(Y ) = Tτ(Z)} , viewed as a vector bundle over Ak−1(G) with respect to the obvious projection of part Z onto Ak−1(G), where ρ : A(G) → TM is the standard anchor of the Lie algebroid and τ : Ak−1(G) → M is the obvious projection. Moreover, the above bundle is canonically a weighted Lie algebroid, a Lie algebroid prolongation in the sense of Popescu and Mart´ ınez.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 16 / 27

slide-113
SLIDE 113

Weighted Lie algebroids out of reductions

Let G ⇒ M be a Lie groupoid and consider the subbundle TkGs ⊂ TkG consisting of all higher order velocities tangent to source-leaves. The bundle Fk = Ak(G) := TkGs

  • M ,

inherits graded bundle structure of degree k as a graded subbundle of

  • TkG. Of course, A = A1(G) can be identified with the Lie algebroid of G.

Theorem

The linearisation of Ak(G) is given as D(Ak(G)) ≃ {(Y , Z) ∈ A(G) ×M TAk−1(G)| ρ(Y ) = Tτ(Z)} , viewed as a vector bundle over Ak−1(G) with respect to the obvious projection of part Z onto Ak−1(G), where ρ : A(G) → TM is the standard anchor of the Lie algebroid and τ : Ak−1(G) → M is the obvious projection. Moreover, the above bundle is canonically a weighted Lie algebroid, a Lie algebroid prolongation in the sense of Popescu and Mart´ ınez.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 16 / 27

slide-114
SLIDE 114

Weighted Lie algebroids out of reductions

Let G ⇒ M be a Lie groupoid and consider the subbundle TkGs ⊂ TkG consisting of all higher order velocities tangent to source-leaves. The bundle Fk = Ak(G) := TkGs

  • M ,

inherits graded bundle structure of degree k as a graded subbundle of

  • TkG. Of course, A = A1(G) can be identified with the Lie algebroid of G.

Theorem

The linearisation of Ak(G) is given as D(Ak(G)) ≃ {(Y , Z) ∈ A(G) ×M TAk−1(G)| ρ(Y ) = Tτ(Z)} , viewed as a vector bundle over Ak−1(G) with respect to the obvious projection of part Z onto Ak−1(G), where ρ : A(G) → TM is the standard anchor of the Lie algebroid and τ : Ak−1(G) → M is the obvious projection. Moreover, the above bundle is canonically a weighted Lie algebroid, a Lie algebroid prolongation in the sense of Popescu and Mart´ ınez.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 16 / 27

slide-115
SLIDE 115

Weighted Lie algebroids out of reductions

Let G ⇒ M be a Lie groupoid and consider the subbundle TkGs ⊂ TkG consisting of all higher order velocities tangent to source-leaves. The bundle Fk = Ak(G) := TkGs

  • M ,

inherits graded bundle structure of degree k as a graded subbundle of

  • TkG. Of course, A = A1(G) can be identified with the Lie algebroid of G.

Theorem

The linearisation of Ak(G) is given as D(Ak(G)) ≃ {(Y , Z) ∈ A(G) ×M TAk−1(G)| ρ(Y ) = Tτ(Z)} , viewed as a vector bundle over Ak−1(G) with respect to the obvious projection of part Z onto Ak−1(G), where ρ : A(G) → TM is the standard anchor of the Lie algebroid and τ : Ak−1(G) → M is the obvious projection. Moreover, the above bundle is canonically a weighted Lie algebroid, a Lie algebroid prolongation in the sense of Popescu and Mart´ ınez.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 16 / 27

slide-116
SLIDE 116

Lagrangian framework for graded bundles

A weighted Lie algebroid on D(Fk) gives the Tulczyjew triple D

  • P(F †

k)

  • Πˆ

ε

✤ ✟✟✟✟✟✟✟✟

❃ TD∗(Fk) ✄✄✄✄✄✄✄✄

❑ ❑ ❑ T∗Fk ✠✠✠✠✠✠✠✠

❀ ❀

ˆ ε

  • Fk

ˆ ρ

  • ✎✎✎✎✎✎✎

TFk−1 ☎☎☎☎☎☎☎☎

Fk

ˆ ρ

  • ✏✏✏✏✏✏✏

✏✏✏✏✏✏✏ ✏✏✏✏✏✏✏

dL

  • λL

♠♠♠♠♠♠

T L

✙ ❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨

Mi(Fk)

❏ ❏ ❏ ❏ ❏

dH

  • D∗(Fk)

▼ ▼

Mi(Fk)

■ ■

Fk−1 Fk−1 Fk−1 Here, the diagram consists of relations, ˆ ε : T∗Fk− −✄T∗D(Fk) → TD∗(Fk), and Mi(Fk) is the so called Mironian of Fk. In the classical case, Mi(TkM) = Tk−1M ×M T∗M. Forget the Hamiltonian side. T L is the Tulczyjew differential and λL the Legendre relation. The fact that we obtain the Euler-Lagrange equations of higher order comes from the vakonomic constraint and the additional gradation.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 17 / 27

slide-117
SLIDE 117

Lagrangian framework for graded bundles

A weighted Lie algebroid on D(Fk) gives the Tulczyjew triple D

  • P(F †

k)

  • Πˆ

ε

✤ ✟✟✟✟✟✟✟✟

❃ TD∗(Fk) ✄✄✄✄✄✄✄✄

❑ ❑ ❑ T∗Fk ✠✠✠✠✠✠✠✠

❀ ❀

ˆ ε

  • Fk

ˆ ρ

  • ✎✎✎✎✎✎✎

TFk−1 ☎☎☎☎☎☎☎☎

Fk

ˆ ρ

  • ✏✏✏✏✏✏✏

✏✏✏✏✏✏✏ ✏✏✏✏✏✏✏

dL

  • λL

♠♠♠♠♠♠

T L

✙ ❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨

Mi(Fk)

❏ ❏ ❏ ❏ ❏

dH

  • D∗(Fk)

▼ ▼

Mi(Fk)

■ ■

Fk−1 Fk−1 Fk−1 Here, the diagram consists of relations, ˆ ε : T∗Fk− −✄T∗D(Fk) → TD∗(Fk), and Mi(Fk) is the so called Mironian of Fk. In the classical case, Mi(TkM) = Tk−1M ×M T∗M. Forget the Hamiltonian side. T L is the Tulczyjew differential and λL the Legendre relation. The fact that we obtain the Euler-Lagrange equations of higher order comes from the vakonomic constraint and the additional gradation.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 17 / 27

slide-118
SLIDE 118

Lagrangian framework for graded bundles

A weighted Lie algebroid on D(Fk) gives the Tulczyjew triple D

  • P(F †

k)

  • Πˆ

ε

✤ ✟✟✟✟✟✟✟✟

❃ TD∗(Fk) ✄✄✄✄✄✄✄✄

❑ ❑ ❑ T∗Fk ✠✠✠✠✠✠✠✠

❀ ❀

ˆ ε

  • Fk

ˆ ρ

  • ✎✎✎✎✎✎✎

TFk−1 ☎☎☎☎☎☎☎☎

Fk

ˆ ρ

  • ✏✏✏✏✏✏✏

✏✏✏✏✏✏✏ ✏✏✏✏✏✏✏

dL

  • λL

♠♠♠♠♠♠

T L

✙ ❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨

Mi(Fk)

❏ ❏ ❏ ❏ ❏

dH

  • D∗(Fk)

▼ ▼

Mi(Fk)

■ ■

Fk−1 Fk−1 Fk−1 Here, the diagram consists of relations, ˆ ε : T∗Fk− −✄T∗D(Fk) → TD∗(Fk), and Mi(Fk) is the so called Mironian of Fk. In the classical case, Mi(TkM) = Tk−1M ×M T∗M. Forget the Hamiltonian side. T L is the Tulczyjew differential and λL the Legendre relation. The fact that we obtain the Euler-Lagrange equations of higher order comes from the vakonomic constraint and the additional gradation.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 17 / 27

slide-119
SLIDE 119

Lagrangian framework for graded bundles

A weighted Lie algebroid on D(Fk) gives the Tulczyjew triple D

  • P(F †

k)

  • Πˆ

ε

✤ ✟✟✟✟✟✟✟✟

❃ TD∗(Fk) ✄✄✄✄✄✄✄✄

❑ ❑ ❑ T∗Fk ✠✠✠✠✠✠✠✠

❀ ❀

ˆ ε

  • Fk

ˆ ρ

  • ✎✎✎✎✎✎✎

TFk−1 ☎☎☎☎☎☎☎☎

Fk

ˆ ρ

  • ✏✏✏✏✏✏✏

✏✏✏✏✏✏✏ ✏✏✏✏✏✏✏

dL

  • λL

♠♠♠♠♠♠

T L

✙ ❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨

Mi(Fk)

❏ ❏ ❏ ❏ ❏

dH

  • D∗(Fk)

▼ ▼

Mi(Fk)

■ ■

Fk−1 Fk−1 Fk−1 Here, the diagram consists of relations, ˆ ε : T∗Fk− −✄T∗D(Fk) → TD∗(Fk), and Mi(Fk) is the so called Mironian of Fk. In the classical case, Mi(TkM) = Tk−1M ×M T∗M. Forget the Hamiltonian side. T L is the Tulczyjew differential and λL the Legendre relation. The fact that we obtain the Euler-Lagrange equations of higher order comes from the vakonomic constraint and the additional gradation.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 17 / 27

slide-120
SLIDE 120

Lagrangian framework for graded bundles

A weighted Lie algebroid on D(Fk) gives the Tulczyjew triple D

  • P(F †

k)

  • Πˆ

ε

✤ ✟✟✟✟✟✟✟✟

❃ TD∗(Fk) ✄✄✄✄✄✄✄✄

❑ ❑ ❑ T∗Fk ✠✠✠✠✠✠✠✠

❀ ❀

ˆ ε

  • Fk

ˆ ρ

  • ✎✎✎✎✎✎✎

TFk−1 ☎☎☎☎☎☎☎☎

Fk

ˆ ρ

  • ✏✏✏✏✏✏✏

✏✏✏✏✏✏✏ ✏✏✏✏✏✏✏

dL

  • λL

♠♠♠♠♠♠

T L

✙ ❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨

Mi(Fk)

❏ ❏ ❏ ❏ ❏

dH

  • D∗(Fk)

▼ ▼

Mi(Fk)

■ ■

Fk−1 Fk−1 Fk−1 Here, the diagram consists of relations, ˆ ε : T∗Fk− −✄T∗D(Fk) → TD∗(Fk), and Mi(Fk) is the so called Mironian of Fk. In the classical case, Mi(TkM) = Tk−1M ×M T∗M. Forget the Hamiltonian side. T L is the Tulczyjew differential and λL the Legendre relation. The fact that we obtain the Euler-Lagrange equations of higher order comes from the vakonomic constraint and the additional gradation.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 17 / 27

slide-121
SLIDE 121

Lagrangian framework for graded bundles

A weighted Lie algebroid on D(Fk) gives the Tulczyjew triple D

  • P(F †

k)

  • Πˆ

ε

✤ ✟✟✟✟✟✟✟✟

❃ TD∗(Fk) ✄✄✄✄✄✄✄✄

❑ ❑ ❑ T∗Fk ✠✠✠✠✠✠✠✠

❀ ❀

ˆ ε

  • Fk

ˆ ρ

  • ✎✎✎✎✎✎✎

TFk−1 ☎☎☎☎☎☎☎☎

Fk

ˆ ρ

  • ✏✏✏✏✏✏✏

✏✏✏✏✏✏✏ ✏✏✏✏✏✏✏

dL

  • λL

♠♠♠♠♠♠

T L

✙ ❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨

Mi(Fk)

❏ ❏ ❏ ❏ ❏

dH

  • D∗(Fk)

▼ ▼

Mi(Fk)

■ ■

Fk−1 Fk−1 Fk−1 Here, the diagram consists of relations, ˆ ε : T∗Fk− −✄T∗D(Fk) → TD∗(Fk), and Mi(Fk) is the so called Mironian of Fk. In the classical case, Mi(TkM) = Tk−1M ×M T∗M. Forget the Hamiltonian side. T L is the Tulczyjew differential and λL the Legendre relation. The fact that we obtain the Euler-Lagrange equations of higher order comes from the vakonomic constraint and the additional gradation.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 17 / 27

slide-122
SLIDE 122

Lagrangian framework for graded bundles

A weighted Lie algebroid on D(Fk) gives the Tulczyjew triple D

  • P(F †

k)

  • Πˆ

ε

✤ ✟✟✟✟✟✟✟✟

❃ TD∗(Fk) ✄✄✄✄✄✄✄✄

❑ ❑ ❑ T∗Fk ✠✠✠✠✠✠✠✠

❀ ❀

ˆ ε

  • Fk

ˆ ρ

  • ✎✎✎✎✎✎✎

TFk−1 ☎☎☎☎☎☎☎☎

Fk

ˆ ρ

  • ✏✏✏✏✏✏✏

✏✏✏✏✏✏✏ ✏✏✏✏✏✏✏

dL

  • λL

♠♠♠♠♠♠

T L

✙ ❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨

Mi(Fk)

❏ ❏ ❏ ❏ ❏

dH

  • D∗(Fk)

▼ ▼

Mi(Fk)

■ ■

Fk−1 Fk−1 Fk−1 Here, the diagram consists of relations, ˆ ε : T∗Fk− −✄T∗D(Fk) → TD∗(Fk), and Mi(Fk) is the so called Mironian of Fk. In the classical case, Mi(TkM) = Tk−1M ×M T∗M. Forget the Hamiltonian side. T L is the Tulczyjew differential and λL the Legendre relation. The fact that we obtain the Euler-Lagrange equations of higher order comes from the vakonomic constraint and the additional gradation.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 17 / 27

slide-123
SLIDE 123

Lagrangian framework for graded bundles

A weighted Lie algebroid on D(Fk) gives the Tulczyjew triple D

  • P(F †

k)

  • Πˆ

ε

✤ ✟✟✟✟✟✟✟✟

❃ TD∗(Fk) ✄✄✄✄✄✄✄✄

❑ ❑ ❑ T∗Fk ✠✠✠✠✠✠✠✠

❀ ❀

ˆ ε

  • Fk

ˆ ρ

  • ✎✎✎✎✎✎✎

TFk−1 ☎☎☎☎☎☎☎☎

Fk

ˆ ρ

  • ✏✏✏✏✏✏✏

✏✏✏✏✏✏✏ ✏✏✏✏✏✏✏

dL

  • λL

♠♠♠♠♠♠

T L

✙ ❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨

Mi(Fk)

❏ ❏ ❏ ❏ ❏

dH

  • D∗(Fk)

▼ ▼

Mi(Fk)

■ ■

Fk−1 Fk−1 Fk−1 Here, the diagram consists of relations, ˆ ε : T∗Fk− −✄T∗D(Fk) → TD∗(Fk), and Mi(Fk) is the so called Mironian of Fk. In the classical case, Mi(TkM) = Tk−1M ×M T∗M. Forget the Hamiltonian side. T L is the Tulczyjew differential and λL the Legendre relation. The fact that we obtain the Euler-Lagrange equations of higher order comes from the vakonomic constraint and the additional gradation.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 17 / 27

slide-124
SLIDE 124

Lagrangian framework for graded bundles

A weighted Lie algebroid on D(Fk) gives the Tulczyjew triple D

  • P(F †

k)

  • Πˆ

ε

✤ ✟✟✟✟✟✟✟✟

❃ TD∗(Fk) ✄✄✄✄✄✄✄✄

❑ ❑ ❑ T∗Fk ✠✠✠✠✠✠✠✠

❀ ❀

ˆ ε

  • Fk

ˆ ρ

  • ✎✎✎✎✎✎✎

TFk−1 ☎☎☎☎☎☎☎☎

Fk

ˆ ρ

  • ✏✏✏✏✏✏✏

✏✏✏✏✏✏✏ ✏✏✏✏✏✏✏

dL

  • λL

♠♠♠♠♠♠

T L

✙ ❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨

Mi(Fk)

❏ ❏ ❏ ❏ ❏

dH

  • D∗(Fk)

▼ ▼

Mi(Fk)

■ ■

Fk−1 Fk−1 Fk−1 Here, the diagram consists of relations, ˆ ε : T∗Fk− −✄T∗D(Fk) → TD∗(Fk), and Mi(Fk) is the so called Mironian of Fk. In the classical case, Mi(TkM) = Tk−1M ×M T∗M. Forget the Hamiltonian side. T L is the Tulczyjew differential and λL the Legendre relation. The fact that we obtain the Euler-Lagrange equations of higher order comes from the vakonomic constraint and the additional gradation.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 17 / 27

slide-125
SLIDE 125

Lagrangian framework for graded bundles

A weighted Lie algebroid on D(Fk) gives the Tulczyjew triple D

  • P(F †

k)

  • Πˆ

ε

✤ ✟✟✟✟✟✟✟✟

❃ TD∗(Fk) ✄✄✄✄✄✄✄✄

❑ ❑ ❑ T∗Fk ✠✠✠✠✠✠✠✠

❀ ❀

ˆ ε

  • Fk

ˆ ρ

  • ✎✎✎✎✎✎✎

TFk−1 ☎☎☎☎☎☎☎☎

Fk

ˆ ρ

  • ✏✏✏✏✏✏✏

✏✏✏✏✏✏✏ ✏✏✏✏✏✏✏

dL

  • λL

♠♠♠♠♠♠

T L

✙ ❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨❨

Mi(Fk)

❏ ❏ ❏ ❏ ❏

dH

  • D∗(Fk)

▼ ▼

Mi(Fk)

■ ■

Fk−1 Fk−1 Fk−1 Here, the diagram consists of relations, ˆ ε : T∗Fk− −✄T∗D(Fk) → TD∗(Fk), and Mi(Fk) is the so called Mironian of Fk. In the classical case, Mi(TkM) = Tk−1M ×M T∗M. Forget the Hamiltonian side. T L is the Tulczyjew differential and λL the Legendre relation. The fact that we obtain the Euler-Lagrange equations of higher order comes from the vakonomic constraint and the additional gradation.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 17 / 27

slide-126
SLIDE 126

Example

Let g be a Lie algebra and put F2 = g2 = g[1] × g[2], with coordinates (xi, zj) on g2 and coordinates (xi, yj, zk) on D(g2) = g[1] × g[1] × g[2]. The embedding ι : g2 ֒ → D(g2) takes the form ι(x, z) = (x, x, z) and the vector bundle projection is τ(x, y, z) = x. The Lie algebroid structure ε : T∗D(g2) → TD∗(g2) reads (x, y, z, α, β, γ) → (x, β, γ, z, ad∗

yβ, α) .

Given a Lagrangian L : g2 → R, the Tulczyjew differential relation T L : g2 → TD∗(g2) is T L(x, z) =

  • x, β, ∂L

∂z (x, z), z, ad∗

xβ, α

  • : α + β = ∂L

∂x (x, z)

  • .

Hence, for the phase dynamics, β = ∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 18 / 27

slide-127
SLIDE 127

Example

Let g be a Lie algebra and put F2 = g2 = g[1] × g[2], with coordinates (xi, zj) on g2 and coordinates (xi, yj, zk) on D(g2) = g[1] × g[1] × g[2]. The embedding ι : g2 ֒ → D(g2) takes the form ι(x, z) = (x, x, z) and the vector bundle projection is τ(x, y, z) = x. The Lie algebroid structure ε : T∗D(g2) → TD∗(g2) reads (x, y, z, α, β, γ) → (x, β, γ, z, ad∗

yβ, α) .

Given a Lagrangian L : g2 → R, the Tulczyjew differential relation T L : g2 → TD∗(g2) is T L(x, z) =

  • x, β, ∂L

∂z (x, z), z, ad∗

xβ, α

  • : α + β = ∂L

∂x (x, z)

  • .

Hence, for the phase dynamics, β = ∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 18 / 27

slide-128
SLIDE 128

Example

Let g be a Lie algebra and put F2 = g2 = g[1] × g[2], with coordinates (xi, zj) on g2 and coordinates (xi, yj, zk) on D(g2) = g[1] × g[1] × g[2]. The embedding ι : g2 ֒ → D(g2) takes the form ι(x, z) = (x, x, z) and the vector bundle projection is τ(x, y, z) = x. The Lie algebroid structure ε : T∗D(g2) → TD∗(g2) reads (x, y, z, α, β, γ) → (x, β, γ, z, ad∗

yβ, α) .

Given a Lagrangian L : g2 → R, the Tulczyjew differential relation T L : g2 → TD∗(g2) is T L(x, z) =

  • x, β, ∂L

∂z (x, z), z, ad∗

xβ, α

  • : α + β = ∂L

∂x (x, z)

  • .

Hence, for the phase dynamics, β = ∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 18 / 27

slide-129
SLIDE 129

Example

Let g be a Lie algebra and put F2 = g2 = g[1] × g[2], with coordinates (xi, zj) on g2 and coordinates (xi, yj, zk) on D(g2) = g[1] × g[1] × g[2]. The embedding ι : g2 ֒ → D(g2) takes the form ι(x, z) = (x, x, z) and the vector bundle projection is τ(x, y, z) = x. The Lie algebroid structure ε : T∗D(g2) → TD∗(g2) reads (x, y, z, α, β, γ) → (x, β, γ, z, ad∗

yβ, α) .

Given a Lagrangian L : g2 → R, the Tulczyjew differential relation T L : g2 → TD∗(g2) is T L(x, z) =

  • x, β, ∂L

∂z (x, z), z, ad∗

xβ, α

  • : α + β = ∂L

∂x (x, z)

  • .

Hence, for the phase dynamics, β = ∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 18 / 27

slide-130
SLIDE 130

Example

Let g be a Lie algebra and put F2 = g2 = g[1] × g[2], with coordinates (xi, zj) on g2 and coordinates (xi, yj, zk) on D(g2) = g[1] × g[1] × g[2]. The embedding ι : g2 ֒ → D(g2) takes the form ι(x, z) = (x, x, z) and the vector bundle projection is τ(x, y, z) = x. The Lie algebroid structure ε : T∗D(g2) → TD∗(g2) reads (x, y, z, α, β, γ) → (x, β, γ, z, ad∗

yβ, α) .

Given a Lagrangian L : g2 → R, the Tulczyjew differential relation T L : g2 → TD∗(g2) is T L(x, z) =

  • x, β, ∂L

∂z (x, z), z, ad∗

xβ, α

  • : α + β = ∂L

∂x (x, z)

  • .

Hence, for the phase dynamics, β = ∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 18 / 27

slide-131
SLIDE 131

Example

Let g be a Lie algebra and put F2 = g2 = g[1] × g[2], with coordinates (xi, zj) on g2 and coordinates (xi, yj, zk) on D(g2) = g[1] × g[1] × g[2]. The embedding ι : g2 ֒ → D(g2) takes the form ι(x, z) = (x, x, z) and the vector bundle projection is τ(x, y, z) = x. The Lie algebroid structure ε : T∗D(g2) → TD∗(g2) reads (x, y, z, α, β, γ) → (x, β, γ, z, ad∗

yβ, α) .

Given a Lagrangian L : g2 → R, the Tulczyjew differential relation T L : g2 → TD∗(g2) is T L(x, z) =

  • x, β, ∂L

∂z (x, z), z, ad∗

xβ, α

  • : α + β = ∂L

∂x (x, z)

  • .

Hence, for the phase dynamics, β = ∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 18 / 27

slide-132
SLIDE 132

Example

This leads to the Euler-Lagrange equations on g2: ˙ x = z , d dt ∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • =

ad∗

x

∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • .

These equations are second order and induce the Euler-Lagrange equations

  • n g which are of order 3:

d dt ∂L ∂x (x, ˙ x) − d dt ∂L ∂z (x, ˙ x)

  • = ad∗

x

∂L ∂x (x, ˙ x) − d dt ∂L ∂z (x, ˙ x)

  • .

For instance, the ‘free’ Lagrangian L(x, z) = 1

2

  • i Ii(zi)2 induces the

equations on g (ck

ij are structure constants, no summation convention):

Ij ... x j =

  • i,k

ck

ij Ikxi ˙

xk . The latter can be viewed as ‘higher Euler equations’.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 19 / 27

slide-133
SLIDE 133

Example

This leads to the Euler-Lagrange equations on g2: ˙ x = z , d dt ∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • =

ad∗

x

∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • .

These equations are second order and induce the Euler-Lagrange equations

  • n g which are of order 3:

d dt ∂L ∂x (x, ˙ x) − d dt ∂L ∂z (x, ˙ x)

  • = ad∗

x

∂L ∂x (x, ˙ x) − d dt ∂L ∂z (x, ˙ x)

  • .

For instance, the ‘free’ Lagrangian L(x, z) = 1

2

  • i Ii(zi)2 induces the

equations on g (ck

ij are structure constants, no summation convention):

Ij ... x j =

  • i,k

ck

ij Ikxi ˙

xk . The latter can be viewed as ‘higher Euler equations’.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 19 / 27

slide-134
SLIDE 134

Example

This leads to the Euler-Lagrange equations on g2: ˙ x = z , d dt ∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • =

ad∗

x

∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • .

These equations are second order and induce the Euler-Lagrange equations

  • n g which are of order 3:

d dt ∂L ∂x (x, ˙ x) − d dt ∂L ∂z (x, ˙ x)

  • = ad∗

x

∂L ∂x (x, ˙ x) − d dt ∂L ∂z (x, ˙ x)

  • .

For instance, the ‘free’ Lagrangian L(x, z) = 1

2

  • i Ii(zi)2 induces the

equations on g (ck

ij are structure constants, no summation convention):

Ij ... x j =

  • i,k

ck

ij Ikxi ˙

xk . The latter can be viewed as ‘higher Euler equations’.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 19 / 27

slide-135
SLIDE 135

Example

This leads to the Euler-Lagrange equations on g2: ˙ x = z , d dt ∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • =

ad∗

x

∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • .

These equations are second order and induce the Euler-Lagrange equations

  • n g which are of order 3:

d dt ∂L ∂x (x, ˙ x) − d dt ∂L ∂z (x, ˙ x)

  • = ad∗

x

∂L ∂x (x, ˙ x) − d dt ∂L ∂z (x, ˙ x)

  • .

For instance, the ‘free’ Lagrangian L(x, z) = 1

2

  • i Ii(zi)2 induces the

equations on g (ck

ij are structure constants, no summation convention):

Ij ... x j =

  • i,k

ck

ij Ikxi ˙

xk . The latter can be viewed as ‘higher Euler equations’.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 19 / 27

slide-136
SLIDE 136

Example

This leads to the Euler-Lagrange equations on g2: ˙ x = z , d dt ∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • =

ad∗

x

∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • .

These equations are second order and induce the Euler-Lagrange equations

  • n g which are of order 3:

d dt ∂L ∂x (x, ˙ x) − d dt ∂L ∂z (x, ˙ x)

  • = ad∗

x

∂L ∂x (x, ˙ x) − d dt ∂L ∂z (x, ˙ x)

  • .

For instance, the ‘free’ Lagrangian L(x, z) = 1

2

  • i Ii(zi)2 induces the

equations on g (ck

ij are structure constants, no summation convention):

Ij ... x j =

  • i,k

ck

ij Ikxi ˙

xk . The latter can be viewed as ‘higher Euler equations’.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 19 / 27

slide-137
SLIDE 137

Example

This leads to the Euler-Lagrange equations on g2: ˙ x = z , d dt ∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • =

ad∗

x

∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • .

These equations are second order and induce the Euler-Lagrange equations

  • n g which are of order 3:

d dt ∂L ∂x (x, ˙ x) − d dt ∂L ∂z (x, ˙ x)

  • = ad∗

x

∂L ∂x (x, ˙ x) − d dt ∂L ∂z (x, ˙ x)

  • .

For instance, the ‘free’ Lagrangian L(x, z) = 1

2

  • i Ii(zi)2 induces the

equations on g (ck

ij are structure constants, no summation convention):

Ij ... x j =

  • i,k

ck

ij Ikxi ˙

xk . The latter can be viewed as ‘higher Euler equations’.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 19 / 27

slide-138
SLIDE 138

Example

This leads to the Euler-Lagrange equations on g2: ˙ x = z , d dt ∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • =

ad∗

x

∂L ∂x (x, z) − d dt ∂L ∂z (x, z)

  • .

These equations are second order and induce the Euler-Lagrange equations

  • n g which are of order 3:

d dt ∂L ∂x (x, ˙ x) − d dt ∂L ∂z (x, ˙ x)

  • = ad∗

x

∂L ∂x (x, ˙ x) − d dt ∂L ∂z (x, ˙ x)

  • .

For instance, the ‘free’ Lagrangian L(x, z) = 1

2

  • i Ii(zi)2 induces the

equations on g (ck

ij are structure constants, no summation convention):

Ij ... x j =

  • i,k

ck

ij Ikxi ˙

xk . The latter can be viewed as ‘higher Euler equations’.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 19 / 27

slide-139
SLIDE 139

Higher order Lagrangian mechanics on Lie algebroids

Let us consider a general Lie groupoid G and a Lagrangian L : Ak → R on Ak = Ak(G). We will refer to such systems as a k-th order Lagrangian system on the Lie algebroid A(G). The relevant diagram here is D ⊂TD∗(Ak(G))

▲ ▲ ▲ ▲ ▲ ▲ ▲ T∗D(Ak(G))

ε

  • ♦♦♦♦♦♦♦♦♦

T∗Ak(G)

r

  • D∗(Ak(G))

TA(G)

D(Ak(G))

ρ

  • Ak(G)

ι

  • dL
  • λL

✗ ❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲

Here, D(Ak(G)) is the corresponding Lie algebroid prolongation, D = ε ◦ r ◦ dL(Ak(G)), and λL is the Legendre relation. Note that we deal with reductions: in the case G is a Lie group, Ak(G) = Tk(G)/G and D(Ak(G)) = TTk−1(G)/G .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 20 / 27

slide-140
SLIDE 140

Higher order Lagrangian mechanics on Lie algebroids

Let us consider a general Lie groupoid G and a Lagrangian L : Ak → R on Ak = Ak(G). We will refer to such systems as a k-th order Lagrangian system on the Lie algebroid A(G). The relevant diagram here is D ⊂TD∗(Ak(G))

▲ ▲ ▲ ▲ ▲ ▲ ▲ T∗D(Ak(G))

ε

  • ♦♦♦♦♦♦♦♦♦

T∗Ak(G)

r

  • D∗(Ak(G))

TA(G)

D(Ak(G))

ρ

  • Ak(G)

ι

  • dL
  • λL

✗ ❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲

Here, D(Ak(G)) is the corresponding Lie algebroid prolongation, D = ε ◦ r ◦ dL(Ak(G)), and λL is the Legendre relation. Note that we deal with reductions: in the case G is a Lie group, Ak(G) = Tk(G)/G and D(Ak(G)) = TTk−1(G)/G .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 20 / 27

slide-141
SLIDE 141

Higher order Lagrangian mechanics on Lie algebroids

Let us consider a general Lie groupoid G and a Lagrangian L : Ak → R on Ak = Ak(G). We will refer to such systems as a k-th order Lagrangian system on the Lie algebroid A(G). The relevant diagram here is D ⊂TD∗(Ak(G))

▲ ▲ ▲ ▲ ▲ ▲ ▲ T∗D(Ak(G))

ε

  • ♦♦♦♦♦♦♦♦♦

T∗Ak(G)

r

  • D∗(Ak(G))

TA(G)

D(Ak(G))

ρ

  • Ak(G)

ι

  • dL
  • λL

✗ ❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲

Here, D(Ak(G)) is the corresponding Lie algebroid prolongation, D = ε ◦ r ◦ dL(Ak(G)), and λL is the Legendre relation. Note that we deal with reductions: in the case G is a Lie group, Ak(G) = Tk(G)/G and D(Ak(G)) = TTk−1(G)/G .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 20 / 27

slide-142
SLIDE 142

Higher order Lagrangian mechanics on Lie algebroids

Let us consider a general Lie groupoid G and a Lagrangian L : Ak → R on Ak = Ak(G). We will refer to such systems as a k-th order Lagrangian system on the Lie algebroid A(G). The relevant diagram here is D ⊂TD∗(Ak(G))

▲ ▲ ▲ ▲ ▲ ▲ ▲ T∗D(Ak(G))

ε

  • ♦♦♦♦♦♦♦♦♦

T∗Ak(G)

r

  • D∗(Ak(G))

TA(G)

D(Ak(G))

ρ

  • Ak(G)

ι

  • dL
  • λL

✗ ❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲

Here, D(Ak(G)) is the corresponding Lie algebroid prolongation, D = ε ◦ r ◦ dL(Ak(G)), and λL is the Legendre relation. Note that we deal with reductions: in the case G is a Lie group, Ak(G) = Tk(G)/G and D(Ak(G)) = TTk−1(G)/G .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 20 / 27

slide-143
SLIDE 143

Higher order Lagrangian mechanics on Lie algebroids

Let us consider a general Lie groupoid G and a Lagrangian L : Ak → R on Ak = Ak(G). We will refer to such systems as a k-th order Lagrangian system on the Lie algebroid A(G). The relevant diagram here is D ⊂TD∗(Ak(G))

▲ ▲ ▲ ▲ ▲ ▲ ▲ T∗D(Ak(G))

ε

  • ♦♦♦♦♦♦♦♦♦

T∗Ak(G)

r

  • D∗(Ak(G))

TA(G)

D(Ak(G))

ρ

  • Ak(G)

ι

  • dL
  • λL

✗ ❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲

Here, D(Ak(G)) is the corresponding Lie algebroid prolongation, D = ε ◦ r ◦ dL(Ak(G)), and λL is the Legendre relation. Note that we deal with reductions: in the case G is a Lie group, Ak(G) = Tk(G)/G and D(Ak(G)) = TTk−1(G)/G .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 20 / 27

slide-144
SLIDE 144

Higher order Lagrangian mechanics on Lie algebroids

Let us consider a general Lie groupoid G and a Lagrangian L : Ak → R on Ak = Ak(G). We will refer to such systems as a k-th order Lagrangian system on the Lie algebroid A(G). The relevant diagram here is D ⊂TD∗(Ak(G))

▲ ▲ ▲ ▲ ▲ ▲ ▲ T∗D(Ak(G))

ε

  • ♦♦♦♦♦♦♦♦♦

T∗Ak(G)

r

  • D∗(Ak(G))

TA(G)

D(Ak(G))

ρ

  • Ak(G)

ι

  • dL
  • λL

✗ ❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲

Here, D(Ak(G)) is the corresponding Lie algebroid prolongation, D = ε ◦ r ◦ dL(Ak(G)), and λL is the Legendre relation. Note that we deal with reductions: in the case G is a Lie group, Ak(G) = Tk(G)/G and D(Ak(G)) = TTk−1(G)/G .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 20 / 27

slide-145
SLIDE 145

Higher order Lagrangian mechanics on Lie algebroids

For instance, using xA as base coordinates, and ya

i as fibre coordinates of

degree i = 1, . . . , k in Ak, extended by the appropriate momenta πj

b of

degree j = 1, . . . , k in D∗(Ak), we get the equations for the Legendre relation in the form (no Lie algebroid structure appears!): kπ1

a = ∂L

∂ya

k

, (k − 1)π2

b =

∂L ∂yb

k−1

− 1 k d dt ∂L ∂yb

k

  • ,

. . . πk

d = ∂L

∂yd

1

− 1 2! d dt ∂L ∂yd

2

  • + 1

3! d2 dt2 ∂L ∂yd

3

  • − · · ·

+(−1)k 1 (k − 1)! dk−2 dtk−2

  • ∂L

∂yd

k−1

  • − (−1)k 1

k! dk−1 dtk−1 ∂L ∂yd

k

  • ,

which we recognise as the Jacobi–Ostrogradski momenta.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 21 / 27

slide-146
SLIDE 146

Higher order Lagrangian mechanics on Lie algebroids

For instance, using xA as base coordinates, and ya

i as fibre coordinates of

degree i = 1, . . . , k in Ak, extended by the appropriate momenta πj

b of

degree j = 1, . . . , k in D∗(Ak), we get the equations for the Legendre relation in the form (no Lie algebroid structure appears!): kπ1

a = ∂L

∂ya

k

, (k − 1)π2

b =

∂L ∂yb

k−1

− 1 k d dt ∂L ∂yb

k

  • ,

. . . πk

d = ∂L

∂yd

1

− 1 2! d dt ∂L ∂yd

2

  • + 1

3! d2 dt2 ∂L ∂yd

3

  • − · · ·

+(−1)k 1 (k − 1)! dk−2 dtk−2

  • ∂L

∂yd

k−1

  • − (−1)k 1

k! dk−1 dtk−1 ∂L ∂yd

k

  • ,

which we recognise as the Jacobi–Ostrogradski momenta.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 21 / 27

slide-147
SLIDE 147

Higher order Lagrangian mechanics on Lie algebroids

For instance, using xA as base coordinates, and ya

i as fibre coordinates of

degree i = 1, . . . , k in Ak, extended by the appropriate momenta πj

b of

degree j = 1, . . . , k in D∗(Ak), we get the equations for the Legendre relation in the form (no Lie algebroid structure appears!): kπ1

a = ∂L

∂ya

k

, (k − 1)π2

b =

∂L ∂yb

k−1

− 1 k d dt ∂L ∂yb

k

  • ,

. . . πk

d = ∂L

∂yd

1

− 1 2! d dt ∂L ∂yd

2

  • + 1

3! d2 dt2 ∂L ∂yd

3

  • − · · ·

+(−1)k 1 (k − 1)! dk−2 dtk−2

  • ∂L

∂yd

k−1

  • − (−1)k 1

k! dk−1 dtk−1 ∂L ∂yd

k

  • ,

which we recognise as the Jacobi–Ostrogradski momenta.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 21 / 27

slide-148
SLIDE 148

Higher order Lagrangian mechanics on Lie algebroids

For instance, using xA as base coordinates, and ya

i as fibre coordinates of

degree i = 1, . . . , k in Ak, extended by the appropriate momenta πj

b of

degree j = 1, . . . , k in D∗(Ak), we get the equations for the Legendre relation in the form (no Lie algebroid structure appears!): kπ1

a = ∂L

∂ya

k

, (k − 1)π2

b =

∂L ∂yb

k−1

− 1 k d dt ∂L ∂yb

k

  • ,

. . . πk

d = ∂L

∂yd

1

− 1 2! d dt ∂L ∂yd

2

  • + 1

3! d2 dt2 ∂L ∂yd

3

  • − · · ·

+(−1)k 1 (k − 1)! dk−2 dtk−2

  • ∂L

∂yd

k−1

  • − (−1)k 1

k! dk−1 dtk−1 ∂L ∂yd

k

  • ,

which we recognise as the Jacobi–Ostrogradski momenta.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 21 / 27

slide-149
SLIDE 149

Higher order Lagrangian mechanics on Lie algebroids

For instance, using xA as base coordinates, and ya

i as fibre coordinates of

degree i = 1, . . . , k in Ak, extended by the appropriate momenta πj

b of

degree j = 1, . . . , k in D∗(Ak), we get the equations for the Legendre relation in the form (no Lie algebroid structure appears!): kπ1

a = ∂L

∂ya

k

, (k − 1)π2

b =

∂L ∂yb

k−1

− 1 k d dt ∂L ∂yb

k

  • ,

. . . πk

d = ∂L

∂yd

1

− 1 2! d dt ∂L ∂yd

2

  • + 1

3! d2 dt2 ∂L ∂yd

3

  • − · · ·

+(−1)k 1 (k − 1)! dk−2 dtk−2

  • ∂L

∂yd

k−1

  • − (−1)k 1

k! dk−1 dtk−1 ∂L ∂yd

k

  • ,

which we recognise as the Jacobi–Ostrogradski momenta.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 21 / 27

slide-150
SLIDE 150

Higher order Lagrangian mechanics on Lie algebroids

For instance, using xA as base coordinates, and ya

i as fibre coordinates of

degree i = 1, . . . , k in Ak, extended by the appropriate momenta πj

b of

degree j = 1, . . . , k in D∗(Ak), we get the equations for the Legendre relation in the form (no Lie algebroid structure appears!): kπ1

a = ∂L

∂ya

k

, (k − 1)π2

b =

∂L ∂yb

k−1

− 1 k d dt ∂L ∂yb

k

  • ,

. . . πk

d = ∂L

∂yd

1

− 1 2! d dt ∂L ∂yd

2

  • + 1

3! d2 dt2 ∂L ∂yd

3

  • − · · ·

+(−1)k 1 (k − 1)! dk−2 dtk−2

  • ∂L

∂yd

k−1

  • − (−1)k 1

k! dk−1 dtk−1 ∂L ∂yd

k

  • ,

which we recognise as the Jacobi–Ostrogradski momenta.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 21 / 27

slide-151
SLIDE 151

Higher order Lagrangian mechanics on Lie algebroids

For instance, using xA as base coordinates, and ya

i as fibre coordinates of

degree i = 1, . . . , k in Ak, extended by the appropriate momenta πj

b of

degree j = 1, . . . , k in D∗(Ak), we get the equations for the Legendre relation in the form (no Lie algebroid structure appears!): kπ1

a = ∂L

∂ya

k

, (k − 1)π2

b =

∂L ∂yb

k−1

− 1 k d dt ∂L ∂yb

k

  • ,

. . . πk

d = ∂L

∂yd

1

− 1 2! d dt ∂L ∂yd

2

  • + 1

3! d2 dt2 ∂L ∂yd

3

  • − · · ·

+(−1)k 1 (k − 1)! dk−2 dtk−2

  • ∂L

∂yd

k−1

  • − (−1)k 1

k! dk−1 dtk−1 ∂L ∂yd

k

  • ,

which we recognise as the Jacobi–Ostrogradski momenta.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 21 / 27

slide-152
SLIDE 152

Higher order Lagrangian mechanics on Lie algebroids

For instance, using xA as base coordinates, and ya

i as fibre coordinates of

degree i = 1, . . . , k in Ak, extended by the appropriate momenta πj

b of

degree j = 1, . . . , k in D∗(Ak), we get the equations for the Legendre relation in the form (no Lie algebroid structure appears!): kπ1

a = ∂L

∂ya

k

, (k − 1)π2

b =

∂L ∂yb

k−1

− 1 k d dt ∂L ∂yb

k

  • ,

. . . πk

d = ∂L

∂yd

1

− 1 2! d dt ∂L ∂yd

2

  • + 1

3! d2 dt2 ∂L ∂yd

3

  • − · · ·

+(−1)k 1 (k − 1)! dk−2 dtk−2

  • ∂L

∂yd

k−1

  • − (−1)k 1

k! dk−1 dtk−1 ∂L ∂yd

k

  • ,

which we recognise as the Jacobi–Ostrogradski momenta.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 21 / 27

slide-153
SLIDE 153

Higher order Lagrangian mechanics on Lie algebroids

The remaining equation for the dynamics is d dt πk

a = ρA a (x) ∂L

∂xA + yb

1 C c ba(x)πk c ,

where ρA

a and C c ba are structure functions of the Lie algebroid A = A(G).

The above equation can then be rewritten as ρA

a (x) ∂L ∂xA =

  • δc

a d dt − yb 1 C c ba(x) ∂L ∂yc

1 − 1

2! d dt

  • ∂L

∂yc

2

  • · · · −(−1)k 1

k! dk−1 dtk−1

  • ∂L

∂yc

k

  • which we define to be the k-th order Euler–Lagrange equations on A(G).

The above higher order algebroid Euler-Lagrange equations are in complete agrement with the ones obtained by J´

  • ´

zwikowski & Rotkiewicz, Colombo & de Diego, as well as Mart´ ınez. We clearly recover the standard higher Euler–Lagrange equations on TkM as a particular example.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 22 / 27

slide-154
SLIDE 154

Higher order Lagrangian mechanics on Lie algebroids

The remaining equation for the dynamics is d dt πk

a = ρA a (x) ∂L

∂xA + yb

1 C c ba(x)πk c ,

where ρA

a and C c ba are structure functions of the Lie algebroid A = A(G).

The above equation can then be rewritten as ρA

a (x) ∂L ∂xA =

  • δc

a d dt − yb 1 C c ba(x) ∂L ∂yc

1 − 1

2! d dt

  • ∂L

∂yc

2

  • · · · −(−1)k 1

k! dk−1 dtk−1

  • ∂L

∂yc

k

  • which we define to be the k-th order Euler–Lagrange equations on A(G).

The above higher order algebroid Euler-Lagrange equations are in complete agrement with the ones obtained by J´

  • ´

zwikowski & Rotkiewicz, Colombo & de Diego, as well as Mart´ ınez. We clearly recover the standard higher Euler–Lagrange equations on TkM as a particular example.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 22 / 27

slide-155
SLIDE 155

Higher order Lagrangian mechanics on Lie algebroids

The remaining equation for the dynamics is d dt πk

a = ρA a (x) ∂L

∂xA + yb

1 C c ba(x)πk c ,

where ρA

a and C c ba are structure functions of the Lie algebroid A = A(G).

The above equation can then be rewritten as ρA

a (x) ∂L ∂xA =

  • δc

a d dt − yb 1 C c ba(x) ∂L ∂yc

1 − 1

2! d dt

  • ∂L

∂yc

2

  • · · · −(−1)k 1

k! dk−1 dtk−1

  • ∂L

∂yc

k

  • which we define to be the k-th order Euler–Lagrange equations on A(G).

The above higher order algebroid Euler-Lagrange equations are in complete agrement with the ones obtained by J´

  • ´

zwikowski & Rotkiewicz, Colombo & de Diego, as well as Mart´ ınez. We clearly recover the standard higher Euler–Lagrange equations on TkM as a particular example.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 22 / 27

slide-156
SLIDE 156

Higher order Lagrangian mechanics on Lie algebroids

The remaining equation for the dynamics is d dt πk

a = ρA a (x) ∂L

∂xA + yb

1 C c ba(x)πk c ,

where ρA

a and C c ba are structure functions of the Lie algebroid A = A(G).

The above equation can then be rewritten as ρA

a (x) ∂L ∂xA =

  • δc

a d dt − yb 1 C c ba(x) ∂L ∂yc

1 − 1

2! d dt

  • ∂L

∂yc

2

  • · · · −(−1)k 1

k! dk−1 dtk−1

  • ∂L

∂yc

k

  • which we define to be the k-th order Euler–Lagrange equations on A(G).

The above higher order algebroid Euler-Lagrange equations are in complete agrement with the ones obtained by J´

  • ´

zwikowski & Rotkiewicz, Colombo & de Diego, as well as Mart´ ınez. We clearly recover the standard higher Euler–Lagrange equations on TkM as a particular example.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 22 / 27

slide-157
SLIDE 157

Higher order Lagrangian mechanics on Lie algebroids

The remaining equation for the dynamics is d dt πk

a = ρA a (x) ∂L

∂xA + yb

1 C c ba(x)πk c ,

where ρA

a and C c ba are structure functions of the Lie algebroid A = A(G).

The above equation can then be rewritten as ρA

a (x) ∂L ∂xA =

  • δc

a d dt − yb 1 C c ba(x) ∂L ∂yc

1 − 1

2! d dt

  • ∂L

∂yc

2

  • · · · −(−1)k 1

k! dk−1 dtk−1

  • ∂L

∂yc

k

  • which we define to be the k-th order Euler–Lagrange equations on A(G).

The above higher order algebroid Euler-Lagrange equations are in complete agrement with the ones obtained by J´

  • ´

zwikowski & Rotkiewicz, Colombo & de Diego, as well as Mart´ ınez. We clearly recover the standard higher Euler–Lagrange equations on TkM as a particular example.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 22 / 27

slide-158
SLIDE 158

Higher order Lagrangian mechanics on Lie algebroids

The remaining equation for the dynamics is d dt πk

a = ρA a (x) ∂L

∂xA + yb

1 C c ba(x)πk c ,

where ρA

a and C c ba are structure functions of the Lie algebroid A = A(G).

The above equation can then be rewritten as ρA

a (x) ∂L ∂xA =

  • δc

a d dt − yb 1 C c ba(x) ∂L ∂yc

1 − 1

2! d dt

  • ∂L

∂yc

2

  • · · · −(−1)k 1

k! dk−1 dtk−1

  • ∂L

∂yc

k

  • which we define to be the k-th order Euler–Lagrange equations on A(G).

The above higher order algebroid Euler-Lagrange equations are in complete agrement with the ones obtained by J´

  • ´

zwikowski & Rotkiewicz, Colombo & de Diego, as well as Mart´ ınez. We clearly recover the standard higher Euler–Lagrange equations on TkM as a particular example.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 22 / 27

slide-159
SLIDE 159

Higher order Lagrangian mechanics on Lie algebroids

The remaining equation for the dynamics is d dt πk

a = ρA a (x) ∂L

∂xA + yb

1 C c ba(x)πk c ,

where ρA

a and C c ba are structure functions of the Lie algebroid A = A(G).

The above equation can then be rewritten as ρA

a (x) ∂L ∂xA =

  • δc

a d dt − yb 1 C c ba(x) ∂L ∂yc

1 − 1

2! d dt

  • ∂L

∂yc

2

  • · · · −(−1)k 1

k! dk−1 dtk−1

  • ∂L

∂yc

k

  • which we define to be the k-th order Euler–Lagrange equations on A(G).

The above higher order algebroid Euler-Lagrange equations are in complete agrement with the ones obtained by J´

  • ´

zwikowski & Rotkiewicz, Colombo & de Diego, as well as Mart´ ınez. We clearly recover the standard higher Euler–Lagrange equations on TkM as a particular example.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 22 / 27

slide-160
SLIDE 160

The tip of a javelin

For instance, let L be the Lagrangian, governing the motion of the tip of a javelin defined on T2R3, L(x, y, z) = 1 2 3

  • i=1

(yi)2 − (zi)2

  • .

We can understand G = R3 here as a commutative Lie group, and since L is G-invariant, we get immediately the reduction to the graded bundle R3[1] × R3[2]. The Euler-Lagrange equations on T2R3, d dt ∂L ∂yi − 1 2 d dt ∂L ∂zi

  • = 0 ,

give in this case dyi dt = 1 2 d2zi dt2 , so the Euler-Lagrange equation on R3 reads d2xi dt2 = 1 2 d4xi dt4 .

Jump to end J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 23 / 27

slide-161
SLIDE 161

The tip of a javelin

For instance, let L be the Lagrangian, governing the motion of the tip of a javelin defined on T2R3, L(x, y, z) = 1 2 3

  • i=1

(yi)2 − (zi)2

  • .

We can understand G = R3 here as a commutative Lie group, and since L is G-invariant, we get immediately the reduction to the graded bundle R3[1] × R3[2]. The Euler-Lagrange equations on T2R3, d dt ∂L ∂yi − 1 2 d dt ∂L ∂zi

  • = 0 ,

give in this case dyi dt = 1 2 d2zi dt2 , so the Euler-Lagrange equation on R3 reads d2xi dt2 = 1 2 d4xi dt4 .

Jump to end J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 23 / 27

slide-162
SLIDE 162

The tip of a javelin

For instance, let L be the Lagrangian, governing the motion of the tip of a javelin defined on T2R3, L(x, y, z) = 1 2 3

  • i=1

(yi)2 − (zi)2

  • .

We can understand G = R3 here as a commutative Lie group, and since L is G-invariant, we get immediately the reduction to the graded bundle R3[1] × R3[2]. The Euler-Lagrange equations on T2R3, d dt ∂L ∂yi − 1 2 d dt ∂L ∂zi

  • = 0 ,

give in this case dyi dt = 1 2 d2zi dt2 , so the Euler-Lagrange equation on R3 reads d2xi dt2 = 1 2 d4xi dt4 .

Jump to end J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 23 / 27

slide-163
SLIDE 163

The tip of a javelin

For instance, let L be the Lagrangian, governing the motion of the tip of a javelin defined on T2R3, L(x, y, z) = 1 2 3

  • i=1

(yi)2 − (zi)2

  • .

We can understand G = R3 here as a commutative Lie group, and since L is G-invariant, we get immediately the reduction to the graded bundle R3[1] × R3[2]. The Euler-Lagrange equations on T2R3, d dt ∂L ∂yi − 1 2 d dt ∂L ∂zi

  • = 0 ,

give in this case dyi dt = 1 2 d2zi dt2 , so the Euler-Lagrange equation on R3 reads d2xi dt2 = 1 2 d4xi dt4 .

Jump to end J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 23 / 27

slide-164
SLIDE 164

The tip of a javelin

For instance, let L be the Lagrangian, governing the motion of the tip of a javelin defined on T2R3, L(x, y, z) = 1 2 3

  • i=1

(yi)2 − (zi)2

  • .

We can understand G = R3 here as a commutative Lie group, and since L is G-invariant, we get immediately the reduction to the graded bundle R3[1] × R3[2]. The Euler-Lagrange equations on T2R3, d dt ∂L ∂yi − 1 2 d dt ∂L ∂zi

  • = 0 ,

give in this case dyi dt = 1 2 d2zi dt2 , so the Euler-Lagrange equation on R3 reads d2xi dt2 = 1 2 d4xi dt4 .

Jump to end J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 23 / 27

slide-165
SLIDE 165

The tip of a javelin

For instance, let L be the Lagrangian, governing the motion of the tip of a javelin defined on T2R3, L(x, y, z) = 1 2 3

  • i=1

(yi)2 − (zi)2

  • .

We can understand G = R3 here as a commutative Lie group, and since L is G-invariant, we get immediately the reduction to the graded bundle R3[1] × R3[2]. The Euler-Lagrange equations on T2R3, d dt ∂L ∂yi − 1 2 d dt ∂L ∂zi

  • = 0 ,

give in this case dyi dt = 1 2 d2zi dt2 , so the Euler-Lagrange equation on R3 reads d2xi dt2 = 1 2 d4xi dt4 .

Jump to end J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 23 / 27

slide-166
SLIDE 166

The tip of a javelin

For instance, let L be the Lagrangian, governing the motion of the tip of a javelin defined on T2R3, L(x, y, z) = 1 2 3

  • i=1

(yi)2 − (zi)2

  • .

We can understand G = R3 here as a commutative Lie group, and since L is G-invariant, we get immediately the reduction to the graded bundle R3[1] × R3[2]. The Euler-Lagrange equations on T2R3, d dt ∂L ∂yi − 1 2 d dt ∂L ∂zi

  • = 0 ,

give in this case dyi dt = 1 2 d2zi dt2 , so the Euler-Lagrange equation on R3 reads d2xi dt2 = 1 2 d4xi dt4 .

Jump to end J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 23 / 27

slide-167
SLIDE 167

The Tulczyjew triple for strings

Using the canonical multisymplectic structure on ∧2T∗M, we get the following Tulczyjew triple for multivector bundles, consisting of double graded bundle morphisms: D

  • T∗ ∧2 T∗M

✌✌✌✌✌✌✌✌✌✌✌

❅ ❅ ❅ ❅

∧2T ∧2 T∗M α2

M

  • β2

M

  • ☞☞☞☞☞☞☞☞☞☞☞

❇ ❇ ❇ ❇ T∗ ∧2 TM ✎✎✎✎✎✎✎✎✎✎✎

❂ ❂ ❂ ❂

∧2TM

✎✎✎✎✎✎✎✎✎✎✎

∧2TM

✌✌✌✌✌✌✌✌✌✌✌

  • ∧2TM

✏✏✏✏✏✏✏✏✏✏✏

dL

  • T L

✗ ❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲

∧2T∗M

❇ ❇ ❇ ❇ ❇

∧2T∗M

❊ ❊ ❊ ❊ ❊

∧2T∗M

❅ ❅ ❅ ❅

M M

  • M

R2

S

  • ∧2TS
  • .

The way of obtaining the implicit phase dynamics D, as a submanifold of ∧2T ∧2 T∗M, from a Lagrangian L : ∧2TM → R (or from a Hamiltonian H : ∧2T∗M → R) is now standard: D = T L(∧2TM).

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 24 / 27

slide-168
SLIDE 168

The Tulczyjew triple for strings

Using the canonical multisymplectic structure on ∧2T∗M, we get the following Tulczyjew triple for multivector bundles, consisting of double graded bundle morphisms: D

  • T∗ ∧2 T∗M

✌✌✌✌✌✌✌✌✌✌✌

❅ ❅ ❅ ❅

∧2T ∧2 T∗M α2

M

  • β2

M

  • ☞☞☞☞☞☞☞☞☞☞☞

❇ ❇ ❇ ❇ T∗ ∧2 TM ✎✎✎✎✎✎✎✎✎✎✎

❂ ❂ ❂ ❂

∧2TM

✎✎✎✎✎✎✎✎✎✎✎

∧2TM

✌✌✌✌✌✌✌✌✌✌✌

  • ∧2TM

✏✏✏✏✏✏✏✏✏✏✏

dL

  • T L

✗ ❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲

∧2T∗M

❇ ❇ ❇ ❇ ❇

∧2T∗M

❊ ❊ ❊ ❊ ❊

∧2T∗M

❅ ❅ ❅ ❅

M M

  • M

R2

S

  • ∧2TS
  • .

The way of obtaining the implicit phase dynamics D, as a submanifold of ∧2T ∧2 T∗M, from a Lagrangian L : ∧2TM → R (or from a Hamiltonian H : ∧2T∗M → R) is now standard: D = T L(∧2TM).

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 24 / 27

slide-169
SLIDE 169

The Tulczyjew triple for strings

Using the canonical multisymplectic structure on ∧2T∗M, we get the following Tulczyjew triple for multivector bundles, consisting of double graded bundle morphisms: D

  • T∗ ∧2 T∗M

✌✌✌✌✌✌✌✌✌✌✌

❅ ❅ ❅ ❅

∧2T ∧2 T∗M α2

M

  • β2

M

  • ☞☞☞☞☞☞☞☞☞☞☞

❇ ❇ ❇ ❇ T∗ ∧2 TM ✎✎✎✎✎✎✎✎✎✎✎

❂ ❂ ❂ ❂

∧2TM

✎✎✎✎✎✎✎✎✎✎✎

∧2TM

✌✌✌✌✌✌✌✌✌✌✌

  • ∧2TM

✏✏✏✏✏✏✏✏✏✏✏

dL

  • T L

✗ ❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲

∧2T∗M

❇ ❇ ❇ ❇ ❇

∧2T∗M

❊ ❊ ❊ ❊ ❊

∧2T∗M

❅ ❅ ❅ ❅

M M

  • M

R2

S

  • ∧2TS
  • .

The way of obtaining the implicit phase dynamics D, as a submanifold of ∧2T ∧2 T∗M, from a Lagrangian L : ∧2TM → R (or from a Hamiltonian H : ∧2T∗M → R) is now standard: D = T L(∧2TM).

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 24 / 27

slide-170
SLIDE 170

The Tulczyjew triple for strings

Using the canonical multisymplectic structure on ∧2T∗M, we get the following Tulczyjew triple for multivector bundles, consisting of double graded bundle morphisms: D

  • T∗ ∧2 T∗M

✌✌✌✌✌✌✌✌✌✌✌

❅ ❅ ❅ ❅

∧2T ∧2 T∗M α2

M

  • β2

M

  • ☞☞☞☞☞☞☞☞☞☞☞

❇ ❇ ❇ ❇ T∗ ∧2 TM ✎✎✎✎✎✎✎✎✎✎✎

❂ ❂ ❂ ❂

∧2TM

✎✎✎✎✎✎✎✎✎✎✎

∧2TM

✌✌✌✌✌✌✌✌✌✌✌

  • ∧2TM

✏✏✏✏✏✏✏✏✏✏✏

dL

  • T L

✗ ❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲❲

∧2T∗M

❇ ❇ ❇ ❇ ❇

∧2T∗M

❊ ❊ ❊ ❊ ❊

∧2T∗M

❅ ❅ ❅ ❅

M M

  • M

R2

S

  • ∧2TS
  • .

The way of obtaining the implicit phase dynamics D, as a submanifold of ∧2T ∧2 T∗M, from a Lagrangian L : ∧2TM → R (or from a Hamiltonian H : ∧2T∗M → R) is now standard: D = T L(∧2TM).

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 24 / 27

slide-171
SLIDE 171

The Euler-Lagrange equations

A surface S : (t, s) → (xσ(t, s)) in M satisfies the Euler-Lagrange equations if the image by dL of its prolongation to ∧2TM, (t, s) →

  • xσ(t, s), ˙

xµν = ∂xµ ∂t ∂xν ∂s − ∂xµ ∂s ∂xν ∂t

  • ,

is α2

M-related to an admissible surface, i.e. the prolongation of a surface

living in the phase space ∧2T∗M to ∧2T ∧2 T∗M. In coordinates, the Euler-Lagrange equations read ˙ xµν = ∂xµ ∂t ∂xν ∂s − ∂xµ ∂s ∂xν ∂t , ∂L ∂xσ = ∂xµ ∂t ∂ ∂s ∂L ∂ ˙ xµσ (t, s)

  • − ∂xµ

∂s ∂ ∂t ∂L ∂ ˙ xµσ (t, s)

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 25 / 27

slide-172
SLIDE 172

The Euler-Lagrange equations

A surface S : (t, s) → (xσ(t, s)) in M satisfies the Euler-Lagrange equations if the image by dL of its prolongation to ∧2TM, (t, s) →

  • xσ(t, s), ˙

xµν = ∂xµ ∂t ∂xν ∂s − ∂xµ ∂s ∂xν ∂t

  • ,

is α2

M-related to an admissible surface, i.e. the prolongation of a surface

living in the phase space ∧2T∗M to ∧2T ∧2 T∗M. In coordinates, the Euler-Lagrange equations read ˙ xµν = ∂xµ ∂t ∂xν ∂s − ∂xµ ∂s ∂xν ∂t , ∂L ∂xσ = ∂xµ ∂t ∂ ∂s ∂L ∂ ˙ xµσ (t, s)

  • − ∂xµ

∂s ∂ ∂t ∂L ∂ ˙ xµσ (t, s)

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 25 / 27

slide-173
SLIDE 173

The Euler-Lagrange equations

A surface S : (t, s) → (xσ(t, s)) in M satisfies the Euler-Lagrange equations if the image by dL of its prolongation to ∧2TM, (t, s) →

  • xσ(t, s), ˙

xµν = ∂xµ ∂t ∂xν ∂s − ∂xµ ∂s ∂xν ∂t

  • ,

is α2

M-related to an admissible surface, i.e. the prolongation of a surface

living in the phase space ∧2T∗M to ∧2T ∧2 T∗M. In coordinates, the Euler-Lagrange equations read ˙ xµν = ∂xµ ∂t ∂xν ∂s − ∂xµ ∂s ∂xν ∂t , ∂L ∂xσ = ∂xµ ∂t ∂ ∂s ∂L ∂ ˙ xµσ (t, s)

  • − ∂xµ

∂s ∂ ∂t ∂L ∂ ˙ xµσ (t, s)

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 25 / 27

slide-174
SLIDE 174

The Euler-Lagrange equations

A surface S : (t, s) → (xσ(t, s)) in M satisfies the Euler-Lagrange equations if the image by dL of its prolongation to ∧2TM, (t, s) →

  • xσ(t, s), ˙

xµν = ∂xµ ∂t ∂xν ∂s − ∂xµ ∂s ∂xν ∂t

  • ,

is α2

M-related to an admissible surface, i.e. the prolongation of a surface

living in the phase space ∧2T∗M to ∧2T ∧2 T∗M. In coordinates, the Euler-Lagrange equations read ˙ xµν = ∂xµ ∂t ∂xν ∂s − ∂xµ ∂s ∂xν ∂t , ∂L ∂xσ = ∂xµ ∂t ∂ ∂s ∂L ∂ ˙ xµσ (t, s)

  • − ∂xµ

∂s ∂ ∂t ∂L ∂ ˙ xµσ (t, s)

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 25 / 27

slide-175
SLIDE 175

The Euler-Lagrange equations

A surface S : (t, s) → (xσ(t, s)) in M satisfies the Euler-Lagrange equations if the image by dL of its prolongation to ∧2TM, (t, s) →

  • xσ(t, s), ˙

xµν = ∂xµ ∂t ∂xν ∂s − ∂xµ ∂s ∂xν ∂t

  • ,

is α2

M-related to an admissible surface, i.e. the prolongation of a surface

living in the phase space ∧2T∗M to ∧2T ∧2 T∗M. In coordinates, the Euler-Lagrange equations read ˙ xµν = ∂xµ ∂t ∂xν ∂s − ∂xµ ∂s ∂xν ∂t , ∂L ∂xσ = ∂xµ ∂t ∂ ∂s ∂L ∂ ˙ xµσ (t, s)

  • − ∂xµ

∂s ∂ ∂t ∂L ∂ ˙ xµσ (t, s)

  • .

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 25 / 27

slide-176
SLIDE 176

Plateau problem

In particular, if M = R3 = {(x1 = x, x2 = y, x3 = z)} with the Euclidean metric, the canonically induced ‘free’ Lagrangian on ∧2TM reads L(xµ, ˙ xκλ) =

  • κ,λ

( ˙ xκλ)2 . The Euler-Lagrange equation for surfaces being graphs (x, y) → (x, y, z(x, y)) provides the well-known equation for minimal surfaces, found already by Lagrange : ∂ ∂x   zx

  • 1 + z2

x + z2 y

  + ∂ ∂y   zy

  • 1 + z2

x + z2 y

  = 0 . In another form: (1 + z2

x )zyy − 2zxzyzxy + (1 + z2 y )zxx = 0 .

Starting with a Lorentz metric, we can obtain analogously the Euler-Lagrange equations for the Nambu-Goto Lagrangian.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 26 / 27

slide-177
SLIDE 177

Plateau problem

In particular, if M = R3 = {(x1 = x, x2 = y, x3 = z)} with the Euclidean metric, the canonically induced ‘free’ Lagrangian on ∧2TM reads L(xµ, ˙ xκλ) =

  • κ,λ

( ˙ xκλ)2 . The Euler-Lagrange equation for surfaces being graphs (x, y) → (x, y, z(x, y)) provides the well-known equation for minimal surfaces, found already by Lagrange : ∂ ∂x   zx

  • 1 + z2

x + z2 y

  + ∂ ∂y   zy

  • 1 + z2

x + z2 y

  = 0 . In another form: (1 + z2

x )zyy − 2zxzyzxy + (1 + z2 y )zxx = 0 .

Starting with a Lorentz metric, we can obtain analogously the Euler-Lagrange equations for the Nambu-Goto Lagrangian.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 26 / 27

slide-178
SLIDE 178

Plateau problem

In particular, if M = R3 = {(x1 = x, x2 = y, x3 = z)} with the Euclidean metric, the canonically induced ‘free’ Lagrangian on ∧2TM reads L(xµ, ˙ xκλ) =

  • κ,λ

( ˙ xκλ)2 . The Euler-Lagrange equation for surfaces being graphs (x, y) → (x, y, z(x, y)) provides the well-known equation for minimal surfaces, found already by Lagrange : ∂ ∂x   zx

  • 1 + z2

x + z2 y

  + ∂ ∂y   zy

  • 1 + z2

x + z2 y

  = 0 . In another form: (1 + z2

x )zyy − 2zxzyzxy + (1 + z2 y )zxx = 0 .

Starting with a Lorentz metric, we can obtain analogously the Euler-Lagrange equations for the Nambu-Goto Lagrangian.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 26 / 27

slide-179
SLIDE 179

Plateau problem

In particular, if M = R3 = {(x1 = x, x2 = y, x3 = z)} with the Euclidean metric, the canonically induced ‘free’ Lagrangian on ∧2TM reads L(xµ, ˙ xκλ) =

  • κ,λ

( ˙ xκλ)2 . The Euler-Lagrange equation for surfaces being graphs (x, y) → (x, y, z(x, y)) provides the well-known equation for minimal surfaces, found already by Lagrange : ∂ ∂x   zx

  • 1 + z2

x + z2 y

  + ∂ ∂y   zy

  • 1 + z2

x + z2 y

  = 0 . In another form: (1 + z2

x )zyy − 2zxzyzxy + (1 + z2 y )zxx = 0 .

Starting with a Lorentz metric, we can obtain analogously the Euler-Lagrange equations for the Nambu-Goto Lagrangian.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 26 / 27

slide-180
SLIDE 180

Plateau problem

In particular, if M = R3 = {(x1 = x, x2 = y, x3 = z)} with the Euclidean metric, the canonically induced ‘free’ Lagrangian on ∧2TM reads L(xµ, ˙ xκλ) =

  • κ,λ

( ˙ xκλ)2 . The Euler-Lagrange equation for surfaces being graphs (x, y) → (x, y, z(x, y)) provides the well-known equation for minimal surfaces, found already by Lagrange : ∂ ∂x   zx

  • 1 + z2

x + z2 y

  + ∂ ∂y   zy

  • 1 + z2

x + z2 y

  = 0 . In another form: (1 + z2

x )zyy − 2zxzyzxy + (1 + z2 y )zxx = 0 .

Starting with a Lorentz metric, we can obtain analogously the Euler-Lagrange equations for the Nambu-Goto Lagrangian.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 26 / 27

slide-181
SLIDE 181

Plateau problem

In particular, if M = R3 = {(x1 = x, x2 = y, x3 = z)} with the Euclidean metric, the canonically induced ‘free’ Lagrangian on ∧2TM reads L(xµ, ˙ xκλ) =

  • κ,λ

( ˙ xκλ)2 . The Euler-Lagrange equation for surfaces being graphs (x, y) → (x, y, z(x, y)) provides the well-known equation for minimal surfaces, found already by Lagrange : ∂ ∂x   zx

  • 1 + z2

x + z2 y

  + ∂ ∂y   zy

  • 1 + z2

x + z2 y

  = 0 . In another form: (1 + z2

x )zyy − 2zxzyzxy + (1 + z2 y )zxx = 0 .

Starting with a Lorentz metric, we can obtain analogously the Euler-Lagrange equations for the Nambu-Goto Lagrangian.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 26 / 27

slide-182
SLIDE 182

Plateau problem

In particular, if M = R3 = {(x1 = x, x2 = y, x3 = z)} with the Euclidean metric, the canonically induced ‘free’ Lagrangian on ∧2TM reads L(xµ, ˙ xκλ) =

  • κ,λ

( ˙ xκλ)2 . The Euler-Lagrange equation for surfaces being graphs (x, y) → (x, y, z(x, y)) provides the well-known equation for minimal surfaces, found already by Lagrange : ∂ ∂x   zx

  • 1 + z2

x + z2 y

  + ∂ ∂y   zy

  • 1 + z2

x + z2 y

  = 0 . In another form: (1 + z2

x )zyy − 2zxzyzxy + (1 + z2 y )zxx = 0 .

Starting with a Lorentz metric, we can obtain analogously the Euler-Lagrange equations for the Nambu-Goto Lagrangian.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 26 / 27

slide-183
SLIDE 183

Plateau problem

In particular, if M = R3 = {(x1 = x, x2 = y, x3 = z)} with the Euclidean metric, the canonically induced ‘free’ Lagrangian on ∧2TM reads L(xµ, ˙ xκλ) =

  • κ,λ

( ˙ xκλ)2 . The Euler-Lagrange equation for surfaces being graphs (x, y) → (x, y, z(x, y)) provides the well-known equation for minimal surfaces, found already by Lagrange : ∂ ∂x   zx

  • 1 + z2

x + z2 y

  + ∂ ∂y   zy

  • 1 + z2

x + z2 y

  = 0 . In another form: (1 + z2

x )zyy − 2zxzyzxy + (1 + z2 y )zxx = 0 .

Starting with a Lorentz metric, we can obtain analogously the Euler-Lagrange equations for the Nambu-Goto Lagrangian.

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 26 / 27

slide-184
SLIDE 184

THANK YOU FOR YOUR ATTENTION!

J.Grabowski (IMPAN) New developments in geometric mechanics B¸ edlewo, 10-16/05/2015 27 / 27