Page 1

arXiv:0707.2819v3 [gr-qc] 30 Mar 2009

Propagation equations for deformable test bodies with microstructure in extended

theories of gravity

Dirk Puetzfeld∗

Institute of Theoretical Astrophysics, University of Oslo, P.O. Box 1029, 0315 Oslo, Norway

Yuri N. Obukhov†

Institute for Theoretical Physics, University of Cologne, Z¨ ulpicher Straße 77, 50937 K¨ oln, Germany‡

(Dated:March 30, 2009)

We derive the equations of motion in metric-affine gravity by making use of the conservation laws

obtained from Noether’s theorem. The results are given in the form of propagation equations for the

multipole decomposition of the matter sources in metric-affine gravity, i.e., the canonical energy-

momentum current and the hypermomentum current. In particular, the propagation equations

allow for a derivation of the equations of motion of test particles in this generalized gravity theory,

and allow for direct identification of the couplings between the matter currents and the gauge

gravitational field strengths of the theory, namely, the curvature, the torsion, and the nonmetricity.

We demonstrate that the possible non-Riemannian spacetime geometry can only be detected with

the help of the test bodies that are formed of matter with microstructure. Ordinary gravitating

matter, i.e., matter without microscopic internal degrees of freedom, can probe only the Riemannian

spacetime geometry. Thereby, we generalize previous results of general relativity and Poincar´ e gauge

theory.

PACS numbers: 04.25.-g; 04.50.+h; 04.20.Fy; 04.20.Cv

Keywords: Approximation methods; Equations of motion; Alternative theories of gravity; Variational prin-

ciples

I.INTRODUCTION

The relation between the field equations and the equations of motion within nonlinear gravitational theories has

been subject to many works. The intimate link between these equations is one of the features of general relativity

which distinguishes it from many other physical theories. The fact that, in contrast to linear field theories, the

equations of motion need not to be postulated separately, but can be derived from the field equations, has been

investigated shortly after the proposal of the theory. From a conceptual standpoint the derivability of the equations

of motion is a very satisfactory result, since it reduces the number of additional assumptions in the theory.1The

earliest accounts of this feature of general relativity can be found in the works of Weyl [2], Eddington [3], as well

as Einstein and Grommer [1]. Nowadays this is customarily addressed as the “problem of motion” in the context of

general relativity and other nonlinear field theories.2

One may distinguish between two conceptually different methods. Both were employed in the derivation of the

equations of motion within the theory of general relativity. One of them goes back to the works of Einstein et al.

[8, 9] and is based on the vacuum field equations of the theory. Within this method matter is modeled in the form of

singularities of the field and only the exterior of bodies is considered. The second method, usually attributed to Fock

∗Electronic address: dirk.puetzfeld@astro.uio.no; URL: http://www.thp.uni-koeln.de/∼dp

†Electronic address: yo@thp.uni-koeln.de

‡Also at Department of Theoretical Physics, Moscow State University, 117234 Moscow, Russia

1The following german quotes are taken from [1] (translation by the authors):

• “[...] Es sieht daher so aus, wie wenn die allgemeine Relativit¨ atstheorie jenen ¨ argerlichen Dualismus bereits siegreich ¨ uberwunden

h¨ atte. [...]”,

“[...] It looks like the general theory of relativity has victoriously overcome this annoying dualism. [...]”.

• “[...] Der hier erzielte Fortschritt liegt aber darin, daß zum ersten Male gezeigt ist, daß eine Feldtheorie eine Theorie des

mechanischen Verhaltens von Diskontinuit¨ aten in sich enthalten kann. [...] ”,

“[...] The progress achieved in this work is that for the first time we have shown that a field theory can contain the theory of the

mechanical behavior of discontinuities. [...]”.

2A historical account of works can also be found in [4, 5, 6, 7].

Page 2

2

[10], makes use of the differential conservation laws of the theory and also allows for a consideration of the interior of

material bodies. In this work we are going to utilize the latter method; i.e., we base our considerations on differential

identities derived from the symmetry of the action via Noether’s theorem.

In addition, we make use of a multipole decomposition of the matter currents. This allows for a systematic study

of the coupling between the matter currents and field strengths of the theory at different orders of approximation.

Multipole methods have been intensively studied in the context of the problem of motion since the early work of

Mathisson [11]. In table I, we provide a corresponding chronological overview.3

TABLE I: Timeline of works which deal with the problem of motion and

multipole approximation schemes.

Year Reference

1923 Weyl [2]

Comment

Mentions the link between the equations of motion (EOM) and the

field equations.

Show that the field equations contain the EOM in GR (for a special

case).

Early investigation regarding the problem of motion, treated as

boundary value problem.

Systematic account of the problem of motion in GR, one of the first

authors who makes use of the δ-function in this context.

Test particle EOM from divergence condition.

Possibly the earliest work utilizing a multipole method in the deriva-

tion of the EOM.

Derivation of the EOM outside of material bodies.

Systematic slow motion approximation.

Gravitational interaction of particles using the multipole method.

Test particle EOM via Gaussian integral transformation.

Derive the geodesic motion of test particles for empty space.

EOM for pole-dipole test particles in GR (see also the later work

[22]).

Derivation of the EOM utilizing a method in the spirit of [10].

Review of the problem of motion in GR.

Relationship of EOM and covariance of a field theory.

1927 Einstein and Grommer [1]

Lanczos [13]

1931 Mathisson [14, 15, 16]

1937 Robertson [17]

Mathisson [11]

1938 Einstein et al. [8, 9]

1939 Fock [10]

1940 Papapetrou [18]

1941 Lanczos [19]

1949 Infeld and Schild [20]

1951 Papapetrou [21]

Papapetrou [23]

1953 Papapetrou [24]

Goldberg [25]

1955 Meister and Papapetrou [26] EOM and coordinate conditions in GR.

1957 Infeld [27]Review of approximation methods, derives EOM using Einstein-

Infeld-Hoffmann (EIH) method, relaxes harmonic coordinate con-

dition, δ-function as source.

1959 Kerr [28, 29]Systematic post-Minkowskian treatment I + II (fast motion

approximation).

Fock [30]Systematic slow motion/weak field approximation.

Tulczyjew [31]Test particle EOM via a simplified version of Mathisson’s method.

1960 Infeld and Plebanski [32]Review of the EIH method.

Kerr [33]Approximation of the quasistatic case, review of three approxima-

tions schemes.

Synge [34]Integralconservation

momentum pseudotensor definition.

1962 Goldberg [5] Review of the problems connected with the EOM in GR and the

EIH method.

Havas and Goldberg [35]Derive single-pole EOM by using Mathisson’s method.

Tulczyjew and Tulczyjew [36] Covariant formulation of a multipole method in GR.

1964 Taub [37]Test particle EOM in a coordinate independent manner using Pa-

papetrou’s method.

Dixon [38]Covariant multipole method for extended test particles in GR.

Havas [39]Generalized version of Mathisson’s method in affine spaces.

1969 Madore [40]EOM for extended bodies using a multipole method which differs

from the one of [21].

1970 Dixon [41, 42]Extended bodies within a multipole formalism.

1973 Liebscher [43, 44]EOM for pole particles in non-Riemannian spaces using the method

in [40], see also [45].

laws, EOMformasscenter,energy-

3An extended version of this table, also including works in the post-Newtonian and post-Minkowskian context, can be found in [12]

Page 3

3

1974 Papapetrou [46]Review of the derivation of the EOM of a single-pole test particle

in GR.

Review of the multipole formalism in GR in the context of extended

bodies.

Generalization of the Papapatrou equations to Poincar´ e gauge

theory.

Multipole method for the derivation of the EOM for extended

bodies.

EOM review.

1979 Dixon [47]

1980 Yasskin and Stoeger [48]

Bailey and Israel [49]

1987 Damour [7]

In this paper, we work out the equations of motion within a multipole formalism for a generalized gravitational

theory known as metric-affine gravity (MAG) [50]. In the theory of general relativity, the mass, or more precisely the

energy-momentum, of matter is the only physical source of the gravitational field. The energy-momentum current

corresponds (via the Noether theorem) to the local translational, or the diffeomorphism, spacetime symmetry. In

MAG, this symmetry is extended to the local affine group that is a semidirect product of translations times the local

linear spacetime symmetry group. Correspondingly, there are additional conserved currents describing microscopic

characteristics of matter that arise as physical sources of the gravitational field. In continuum mechanics [51, 52,

53, 54, 55, 56], such matter is described as a medium with microstructure. In physical terms this means that the

elements of a material continuum have internal degrees of freedom such as spin, dilation, and shear. The three latter

microscopic sources are represented in MAG by the irreducible parts (that correspond to the Lorentz, dilational

and shear-deformational subgroups of the general linear group) of the hypermomentum current. Fluid models with

microstructure were extensively studied within different gravity theories (including MAG), see, e.g., [57, 58, 59, 60, 61].

The metric-affine theory naturally generalizes the Poincar´ e gravity theory [62, 63] in which the mass (energy-

momentum) and spin are the sources of the gravitational field. The geometry that arises on the spacetime manifold is

non-Riemannian, it is known as the Riemann-Cartan geometry with curvature and torsion. In MAG, this geometrical

structure is further extended to the metric-affine spacetime with curvature, torsion, and nonmetricity. The resulting

general scheme of MAG embeds not only Poincar´ e gravity, but also a wide spectrum of gauge gravitational models

based on the conformal, Weyl, de Sitter, and other spacetime symmetry groups (for an overview, see [50], for example).

This fact makes the analysis of the equations of motion in MAG especially interesting, with possible direct physical

applications for all the gravitational models mentioned.

The energy-momentum current and the hypermomentum current (spin + dilaton + shear charge) are the sources

of the gravitational field in MAG. Accordingly, test bodies that are formed of matter with microstructure have

two kinds of physical properties which determine their dynamics in a curved spacetime. The properties of the first

type have microscopic origin; they arise directly from the fact that the elements of a medium have internal degrees of

freedom (microstructure). The properties of the second type are essentially macroscopic; they arise from the collective

dynamics of matter elements characterized by mass (energy) and momentum. More exact definitions will be given

later, but the qualitative picture is as follows. The averaging of the microscopic hypermomentum current yields the

integrated spin, dilaton, and shear charge of a test body. In addition, the averaging of the energy-momentum and of

its multipole moments gives rise to the orbital integrated momenta. In Poincar´ e gravity, there is only one relevant

first moment, namely, the orbital angular momentum. It describes the behavior of a test particle as a rigid body,

i.e., its rotation. In metric-affine gravity, one finds, in addition, the orbital moments that describe deformations of

body. These are the orbital dilation momentum (that describes isotropic volume expansion) and the orbital shear

momentum (that determines the anisotropic deformations with fixed volume). The three together (orbital angular

momentum, orbital dilation momentum, and orbital shear momentum) comprise the generalized integrated orbital

momentum. In this paper, we compare the gravitational interaction of the integrated hypermomentum to that of the

integrated orbital momentum of a rotating and deformable test body. Thereby, we generalize the previous analysis

[48] in which the effects of the integrated spin were compared to the effects of the orbital angular momentum of a

rotating rigid test body.

The paper is organized as follows. In section II we recall some basic facts about the gravity theory under consid-

eration, namely, metric-affine gravity. This is followed by a discussion of the conservation laws within this theory in

section III which form the basis for the derivation of the equations of motion. We then work out the explicit form

of the propagation equations in sections IV and V. In section VI we provide some relations between the different

definitions of momenta within the multipole formalism. We discuss our findings in section VII and present an outlook

on the open questions within this field. Our notation and conventions are summarized in appendix A. A table with

the dimensions of all quantities appearing throughout the work can be found in appendix B.

Page 4

4

II.METRIC-AFFINE GRAVITY

Metric-affine gravity represents a gauge-theoretical formulation of a theory of gravitation which is based on the

general affine group A(4,R), i.e., the semidirect product of the four-dimensional translation group R4and the general

linear group GL(4,R). For a review of the theory see [50, 64], and references therein. In such a theory, besides

the usual “weak” Newton-Einstein–type gravity, described by the metric of spacetime, additional “strong” gravity

pieces will arise that are supposed to be mediated by additional degrees of freedom related to the independent linear

connection Γαβ. Alternatively, the strong gravity pieces can also be expressed in terms4of the nonmetricity Qαβand

the torsion Tα. The propagating modes related to the new degrees of freedom are expected to manifest themselves

in the non-Riemannian pieces of the curvature Rαβ. The existence of such modes certainly depends on the choice

of the dynamical scheme, or in technical terms, on the choice of the Lagrangian. The simplest generalization of the

linear Hilbert-Einstein Lagrangian leads to a model with contact interaction. However, quadratic Yang-Mills–type

Lagrangians describe a wide spectrum of non-Riemannian propagating gravitational modes. This is revealed, for

example, by studies of generalized gravitational waves in models with torsion [65, 66, 67, 68, 69, 70, 71] and in models

with torsion and nonmetricity [72, 73, 74, 75, 76, 77, 78, 79, 80].

In a Lagrangian framework one usually considers the geometrical “potentials” (metric gαβ, coframe 1-form ϑα,

connection 1-form Γαβ) to be minimally coupled to matter fields, collectively called ψ, such that the total Lagrangian,

i.e., the geometrical and the matter part, is given by

Ltot= L?gαβ,ϑα,Qαβ,Tα,Rαβ?+ Lmat(gαβ,ϑα,ψ,Dψ).

Here D = d+ℓαβΓαβ, with ℓαβdenoting the generators of the linear transformations (namely, δψ = εβαℓαβψ, where

εβαare the infinitesimal parameters). With the following general definitions for the gauge field momenta

(1)

Mαβ:= −2

∂L

∂Qαβ,Hα:= −∂L

∂Tα,Hαβ:= −

∂L

∂Rαβ,(2)

the field equations of metric-affine gravity take the form

(δ/δgαβ)

(δ/δϑα)

?δ/δΓαβ?

(matter)

DMαβ− mαβ= σαβ,

DHα− Eα= Σα,

DHαβ− Eαβ= ∆αβ,

δL

δψ= 0.

(3)

(4)

(5)

(6)

On the right-hand side (rhs) of the field equations we have the physical sources: the metrical energy-momentum σαβ,

the canonical energy-momentum Σα, and the canonical hypermomentum ∆αβcurrents of the matter fields

σαβ:= 2δLmat

δgαβ

,Σα:=δLmat

δϑα,∆αβ:=δLmat

δΓαβ.(7)

On the left-hand side (lhs) there are typical Yang-Mills–like terms governing the gauge gravitational fields, and the

corresponding terms that describe the currents of the gauge fields themselves that arise due to the nonlinearity of

the theory. The metrical energy-momentum, the canonical energy-momentum, and the canonical hypermomentum

currents of the gauge gravitational fields are introduced by

mαβ:= 2∂L

∂gαβ,Eα:=

∂L

∂ϑα,Eαβ:=

∂L

∂Γαβ.(8)

MAG has a wide gauge symmetry group. With the help of the Noether theorems for the diffeomorphism symmetry and

for the local linear symmetry, one can verify that [provided the matter field equations (6) are fulfilled] the following

identities hold:

Σα = eα⌋Lmat− (eα⌋Dψ) ∧∂Lmat

∂Dψ

− (eα⌋ψ) ∧∂Lmat

∂ψ

,(9)

4Please see appendix A on page 22 for the definitions of the objects in this section and a short summary of our conventions.

Page 5

5

Eα = eα⌋L + (eα⌋Tβ) ∧ Hβ+ (eα⌋Rβγ) ∧ Hβγ+1

Eαβ = −ϑα∧ Hβ− Mαβ,

∆αβ = (ℓαβψ) ∧∂Lmat

?eα⌋Tβ?∧ Σβ−1

D∆αβ = gβγσαγ− ϑα∧ Σβ.

The gauge symmetry and the corresponding Noether identities play an essential role in MAG. The most important

result is as follows: It can be shown that, by means of (10)-(14), the field equation (3) is redundant. It is a consequence

of the two other MAG field equations (4) and (5) and of the Noether identities. The explanation is straightforward:

One can use the local linear transformations of the frames to “gauge away” the metric gαβby making it equal to the

constant Minkowski metric diag(1,−1,−1,−1) everywhere on the spacetime manifold. After doing this, equation (3)

is trivially solved, and one needs to solve only the remaining equations (4) and (5) to determine the coframe ϑαand

connection Γαβ.

There are many nontrivial exact solutions for different MAG models ranging from black holes, gravitational waves,

to cosmological models known in the literature. Nearly all of the corresponding references can be found in the works

[50, 81, 82, 83].

2(eα⌋Qβγ)Mβγ,(10)

(11)

∂Dψ,

(12)

DΣα =

2(eα⌋Qβγ)σβγ+ (eα⌋Rβγ) ∧ ∆βγ, (13)

(14)

III. CONSERVATION LAWS

An up-to-date discussion of the conservation laws within metric-affine gravity can be found in the recent work

[84]. In the following sections IIIA-IIIC we recall the conservation laws for the canonical energy-momentum and

hypermomentum. These conservation laws serve as starting point for our subsequent derivation of the propagation

equations for the multipole moments of the matter currents. In IIIC we make contact with Poincar´ e gauge theory,

which represents the special case of metric-affine gravity for which the distorsion, i.e., the difference between the full

and the metric-compatible connection, reduces to the antisymmetric contortion, and the hypermomentum reduces to

the spin current.

A.Energy-momentum conservation

The Noether theorem for the diffeomorphism invariance of the matter action yields the conservation law of the

energy-momentum current

?

{ }

D

?Σα− ∆γβeα⌋Nγβ?≡eα⌋

{ }

Rγβ−

{ }

? LαNγβ

?

∧ ∆γβ. (15)

Here

After we substitute the components from (A5)-(A10), we finally find the tensor form of the conservation law (15):

{ }

? Lξ= ξ⌋

{}

D +

{}

Dξ⌋ is the (Riemannian) covariant Lie derivative.

{ }

∇j

?Tij− Nikl∆klj?=?{}

Rijkl−

{}

∇iNjkl

?∆klj.(16)

This can be identically rewritten as

{}

∇jTij=?Rijkl∆klj+ Nikl

{}

∇j∆klj,(17)

where we denoted

?Rijkl:=

{ }

Rijkl−

{ }

∇iNjkl+

{}

∇jNikl. (18)

Page 6

6

B.Hypermomentum conservation

The Noether theorem for the local GL(4,R)-invariance of MAG yields (on the mass shell, i.e., when the matter

satisfies the field equations):

D∆αβ+ ϑα∧ Σβ− σαβ= 0.(19)

Here the last term describes the metrical energy-momentum 4-form defined in equation (7). By the introduction of

local coordinates for the corresponding components,

σαβ= tαβη,(20)

we can rewrite the Noether identity (19) in tensorial form:

{}

∇j∆klj− Nijk∆jli+ Njli∆kij+ Tlk− tkl= 0.(21)

Taking the antisymmetric part, we find:

{ }

∇j∆[kl]j= Nij[k∆|j|l]i+ Nj[k|i|∆l]ij+ T[kl]. (22)

C.Recovering Poincar´ e gauge theory

The case of the Poincar´ e gauge theory is recovered when the difference of the connections reduces to the anti-

symmetric contortion Nαβ= Kαβ= K[αβ], whereas the hypermomentum reduces to the antisymmetric spin current

∆αβ= ταβ= τ[αβ].

With the help of (22), we then immediately find

Kikl

{}

∇jτklj= KiklTkl+ (KinlKjkn− KjnlKikn)τklj. (23)

Substituting this into (17), and rearranging the rhs, we have

{}

∇jTij= Rijklτklj+ KiklTkl.(24)

Here the total Riemann-Cartan curvature is recovered in the first term on the rhs:

Rijkl=

{}

Rijkl−

{}

∇iKjkl+

{}

∇jKikl+ KinlKjkn− KjnlKikn,(25)

in complete agreement with (A2).

Now, writing down explicitly the Riemannian covariant derivative, we recast (24) into

?√−gTij?=√−g?ΓijkTkj+ Rijklτklj?.

Here, the first term on the rhs contains the full Riemann-Cartan connection, Γijk=

It is also possible to write the conservation law in a different form. By raising the index i, we then straightforwardly

can recast (24) into

??

∂j

(26)

{}

Γ ijk− Kijk, cf. with (A1).

∂j

?√−gTij?=√−gKikl−

{}

Γ kli

?

Tkl+ Rijklτklj

?

. (27)

Thus, equation (42) of [48] is correct, it coincides with (27). However, one should be careful since the position of

indices in the definitions of the connection, torsion, contortion, and curvature is different from our conventions. Note

also that the spin in Yasskin and Stoeger is defined with the

it with our definition (A4). It is satisfactory to see that our computations regarding the conservation laws are in

complete agreement with those of Yasskin and Stoeger in [48].

1

2factor, see their definition (8) in [48], and compare

Page 7

7

FIG. 1: Sketch of the hypersurface Σ, i.e., the world tube of the test particle.

parametrized by Ya. Coordinates within the world tube with respect to a coordinate system centered on Yaare labeled by xa.

A continuous curve through the tube is

IV.PROPAGATION EQUATIONS

Let us switch to a notation which is close to the one in [48]. It turns out that (17) is more appropriate to bring the

energy-momentum conservation equation into a form analogous to the result (42) in [48]. By raising one index and

explicitly rewriting5the covariant derivative in the first term of (17), we obtain

?Tij,j=?Rijkl?∆klj−

{}

Γ kji?T(kj)+ Nikl

{}

∇j?∆klj.(28)

Furthermore, the hypermomentum conservation equation in (21) takes the form

?∆klj,j= Nmjk?∆jlm−

{}

Γ mjk?∆mlj−

{}

Γ mjl?∆k(mj)− Njlm?∆kmj−?Tlk+?tkl.(29)

By using (21) in (17), we can also obtain an alternative version of (28), which has a very similar structure compared

to (42) in [48]:

?Tij,j = ?Rijkl?∆klj+ NiklNajk?∆jla− NiklNjla?∆kaj− Nikl?Tlk−

{ }

Γ kji?T(kj)+ Nikl?tkl

⇔?Tij,j = Rijkl?∆klj− Nikl?Tlk−

Note that in the last equation Rijkl represents the full curvature. The structure of equation (30) is very similar to

(42) in [48]. In the following we are going to derive the propagation equations for the integrated moments following

from the conservation equations (29) and (30).

{}

Γ kji?T(kj)+ Nikl?tkl. (30)

5Remember

{ }

∇j

?

Sij+ Aij?

=

1

√−g

?√−g?

Sij+ Aij??

,j+

{ }

ΓkjiSkj, where Sijdenotes the symmetric and Aijdenotes the antisym-

metric part of a quantity with two indices.

Page 8

8

A.Lemma: Derivative of the integrated moments

The following relation, cf. (41) in [48], between the time derivative of the multipole expansion of a current also

holds within metric-affine gravity

?

j=1

i=1

j=1, j?=i

d

dt

n

?

δxbj

JA0=

n

?

ρbia

?

n

?

δxbj

JAa+

?

n

?

j=1

δxbj

JAa,a.(31)

Here JAdenotes the density of a matter current, in our case?∆klj,?Tij, or?tkl. Additionally, we have δxa:= xa− Ya,

the last index of the corresponding matter current, e.g., JA0→?Ti0. In (31), and in the following, integrals are taken

?

and ρba= δxb,a= δb

a− vbδ0

a= δb

a− δb

0δ0

a= δb

αδα

afor the spatial projector. The upper-index of JAais associated with

over a 3-dimensional slice Σ(t), at a time t, of the world tube of a test body. We use the condensed notation

?

f =

Σ(t)

f(x)d3x.

B.Conservation equations integrated

With the help of (31), we derive the integrated version of the conservation equations (30):

??

α=1

d

dt

??

n

?

n

?

α=1

δxbα

?

??

?Ti0=

Rijkl?∆klj− Nikl?Tlk−

n

?

β=1

?

n

?

α=1,α?=β

δxbα

?Tibβ− vbβ

Γ kji?T(kj)+ Nikl?tkl

?

?

n

?

α=1,α?=β

δxbα

?Ti0

+δxbα

{}

,

and (29)

d

dt

??

??

n

?

n

?

α=1

δxbα

?

??

?∆kl0=

Nmjk?∆jlm−

n

?

β=1

?

Γ mjk?∆mlj−

n

?

α=1,α?=β

δxbα

?∆klbβ− vbβ

{ }

Γ mjl?∆k(mj)− Njlm?∆kmj−?Tlk+?tkl

?

n

?

α=1,α?=β

δxbα

?∆kl0

+

α=1

δxbα

{ }

?

.

With the introduction of new names for the integrated moments

∆b1···bnijk

: =

??

??

??

n

?

n

?

n

?

α=1

δxbα

?

?

?

?∆ijk,

?Tij,

?tij,

T

b1···bnij

: =

α=1

δxbα

tb1···bnij

: =

α=1

δxbα

(32)

the integrated conservation laws take the following form:6

d

dtT

b1···bni0=

n

?

β=1

?

T

b1···ˇbβ···bnibβ− vbβT

b1···ˇbβ···bni0?

6Note that we use an inverted circumflex, e.g.,ˇbβ, to indicate that an index is omitted from a list.

Page 9

9

+

??

n

?

α=1

δxbα

??

n

?

??

Rijkl?∆klj− Nikl?Tlk−

∆

{ }

Γ kji?T(kj)+ Nikl?tkl

?

,(33)

d

dt∆

??

b1···bnkl0=

β=1

?

b1···ˇbβ···bnklbβ− vbβ∆

b1···ˇbβ···bnkl0?

+

n

?

α=1

δxbα

Nmjk?∆jlm−

{ }

Γ mjk?∆mlj−

{ }

Γ mjl?∆k(mj)− Njlm?∆kmj−?Tlk+?tkl

?

.(34)

Equations (33) and (34) may be compared to (51) and (52) in [48].

C.Propagation equations for pole-dipole particles

Let us now proceed along the lines of [48] and derive the propagation equations for pole-dipole particles by using (33)

and (34). Here we investigate the case in which the following moments are nonvanishing: ∆

– i.e., we only take into account a pole contribution from the hypermomentum; the canonical energy-momentum and

symmetric energy-momentum are considered to contribute at the pole as well as at the dipole level. This assumption

is in accordance with the treatment in [48], in which only pole contributions of the spin current were considered. Let

us expand the geometrical quantities around the worldline Y (t) of the test particle, cf. figure 1, into a power series in

δxa= xa− Ya. We have

Rijkl

????

The general form of the integrated conservation laws (33) and (34) then yields the following set of propagation

equations:

ijk,T

ij,T

ijk,tij, and tijk

??

??

x= Rijkl

??

Y

Y+ δxaNikl,a

Y+ δxaRijkl,a

+ δxa{ }

??

Y

Y+ ···,

{ }

Γ ijk

x

x= Nikl

=

{}

Γ ijk

????

Γ ijk,a

????

+ ···,

Nikl

????

Y+ ···.(35)

d

dtT

i0

= Rijkl∆

klj− NiklT

lk− Nikl,aT

alk−

{}

Γ kjiT

(kj)−

{}

Γ kji,aT

a(kj)

+Nikltkl+ Nikl,atakl,(36)

d

dtT

ai0

= T

ia− vaT

bia+ T

i0− NiklT

aib− vaT

jlm−

kla− va∆

alk−

{}

Γ kjiT

a(kj)+ Nikltakl,(37)

0 = T

bi0− vbT

Γ mjk∆

ai0, (38)

d

dt∆

kl0

= Nmjk∆

{}

mlj−

alk+ takl.

{}

Γ mjl∆

k(mj)− Njlm∆

k

mj− T

lk+ tkl,(39)

0 = ∆

kl0− T(40)

Here we suppressed the dependencies on the points at which certain quantities are evaluated. The set (36)-(40)

represents the generalization of the propagation equations (63)-(67) in [48] to metric-affine gravity.

V. ALTERNATIVE FORM OF THE PROPAGATION EQUATIONS

It was pointed out by several authors, see also page 2086 in [48], that the form of the propagation equations depends

on the definition of the integrated moments, in particular, the index position in the set of equations (32). Of course

ambiguities emerge due to the integration process and the fact that the metric is not a constant. In the previous

section we used the index positions which match the ones used in [48]; this allows for a direct comparison of their

propagation equations with our result in metric-affine gravity. Since there is a priori no way to tell which index

position in the integrated moments is the more physical one, we are also going to derive an alternative version of the

propagation equations, in which integrated moments with mixed indices are used.

From a formal standpoint, the definition with mixed indices may be favored over the definition with upper indices.

Geometrically, the momentum should always be a covector, i.e., it should have a lower-index. This becomes imme-

diately clear if we recall some basic facts from classical mechanics. The velocity is a vector (with an upper-index),

Page 10

10

va= ˙ qa. Then, the momentum is, by definition, pa := ∂L/∂va– which obviously is a covector. Hence, from this

standpoint it appears plausible to consider the choice

?

as definition for the momentum. In the following, we are going to work out an alternative set of propagation equations,

which are based on the definitions with mixed indices.

Once again, we start by rewriting the conservation equations for the canonical energy-momentum current (17) and

hypermomentum current (21), which take the following form

Pa=

?Ta0,

?Tij,j = Rijkl?∆klj+ Γijk?Tkj+ Nijk?tjk,(41)

?∆klj,j = Γjlm?∆kmj− Γmjk?∆jlm−?Tlk+?tkl.(42)

Note that Γijkrepresents the full connection, the last two equations should be compared to (42) and (43) in [48].

Apart from the index positions, equations (41) and (42) are completely equivalent to (30) and (29).

A.Conservation equations integrated

With the help of (31), we derive the integrated version of the conservation equations (41):

??

α=1

d

dt

??

n

?

n

?

α=1

δxbα

?

??

?Ti0=

Rijkl?∆klj+ Γijk?Tkj+ Nijk?tjk

n

?

β=1

?

n

?

α=1,α?=β

δxbα

?Tibβ− vbβ

?

?

n

?

α=1,α?=β

δxbα

?Ti0

+

δxbα

,

and (29)

d

dt

??

??

n

?

n

?

α=1

δxbα

?

??

?∆kl0=

Γjlm?∆kmj− Γmjk?∆jlm−?Tlk+?tkl

n

?

β=1

?

n

?

α=1,α?=β

δxbα

?∆klbβ− vbβ

?

?

n

?

α=1,α?=β

δxbα

?∆kl0

+

α=1

δxbα

.

Now we introduce the integrated moments with mixed index positions. Note that we use an underline (lower-index

position) to distinguish these definitions from the overlined (upper-index position) quantities in (32)

??

??

??

∆b1···bnijk: =

n

?

n

?

n

?

α=1

δxbα

?

?

?

?∆ijk,

?Tij,

?tij.

Tb1···bnij: =

α=1

δxbα

tb1···bnij : =

α=1

δxbα

(43)

With these definitions the integrated conservation laws take the following form

d

dtTb1···bni0=

??

α=1

n

?

??

β=1

?

Tb1···ˇbβ···bnibβ− vbβTb1···ˇbβ···bni0?

+

n

?

δxbα

Rijkl?∆klj+ Γijk?Tkj+ Nijk?tjk

?

,(44)

Page 11

11

d

dt∆b1···bnkl0=

??

α=1

n

?

??

β=1

?

∆b1···ˇbβ···bnklbβ− vbβ∆b1···ˇbβ···bnkl0?

+

n

?

δxbα

Γjlm?∆kmj− Γmjk?∆jlm−?Tlk+?tkl

?

. (45)

Equations (44) and (45) should be compared to (33) and (34), as well as to equations (51) and (52) in [48].

B.Propagation equations for pole-dipole particles

Finally, we derive the propagation equations for pole-dipole particles by using (44) and (45). Again we investigate

the case in which the following moments are nonvanishing: ∆ijk,Tij,Tijk,tij, and tijk. The expansion of geometrical

quantities around the worldline Y (t) of the test particle, cf. figure 1, into a power series in δxa= xa− Ya, reads

Rijkl??

Nijk??

The general form of the integrated conservation laws (44) and (45) then yields the following set of propagation

equations:

x= Rijkl??

x= Nijk??

Y+ δxaRijkl,a

Y+ δxaΓijk,a

Y+ δxaNijk,a

??

Y+ ···.

Y+ ···,

Y+ ···,

??

Γijk??

x= Γijk????

(46)

d

dtTi

d

dtTai0= Ti

0 = Tbia+ Taib− vaTbi0− vbTai0,

d

dt∆kl0= Γjlm∆kmj− Γmjk∆jlm− Tl

0 = ∆kla− va∆kl0− Talk+ takl.

0= Rijkl∆klj+ ΓijkTk

j+ Γijk,aTakj+ Nijktjk+ Nijk,atajk,(47)

a− vaTi

0+ ΓijkTakj+ Nijktajk,(48)

(49)

k+ tkl,(50)

(51)

Again we suppressed the dependencies on the points at which certain quantities are evaluated. The set (47)-(51)

represents the generalization of the propagation equations (63)-(67) in [48] to metric-affine gravity, now with the

mixed index convention. The above set should be compared to our result in (36)-(40).

C. Rewriting the propagation equations ` a la Yasskin and Stoeger

Now let us rewrite the propagation equations of metric-affine gravity (47)-(51) in a form which closely resembles

the main theorem of Yasskin and Stoeger in Poincar´ e gauge theory, i.e., equations (53)-(58) in [48]. We start with

the following identity which holds because of the definition of the projector ρab:

∆bca= va∆bc0+ ρak∆bck. (52)

Using this relation, the last one of the propagation equations (51) takes the form

Talk− takl= ρab∆klb.(53)

This equation may be compared to equation (68) in [48]. Again with the help of (52) we can rewrite (50) as follows:

?

where

tkl− Tl

k= ∇v∆kl0+Γjmk∆mlb− Γjlm∆kmb?

ρjb,(54)

∇v∆kl0:=d

dt∆kl0+ vmΓmjk∆jl0− vmΓmlj∆kj0.(55)

Page 12

12

Equation (54) should be compared to equation (69) in [48]. Proceeding along similar lines as in [21, 48], we are now

going to cyclically permute the indices in (49) twice, resulting in

0 = Tbia+ Taib− vaTbi0− vbTai0,

0 = Tiab+ Tbai− vbTia0− viTba0,

0 = Tabi+ Tiba− viTab0− vaTib0.

(56)

(57)

(58)

Then adding (56) and (57) and subtracting (58) yields

0 = Tb(ai)− Ta[bi]− Ti[ba]− vaT[bi]0− vbT(ai)0− viT[ba]0.(59)

This equation should be compared to (72) in [48]. We proceed with the following identity:

2T(ab)0= 2Ta[b0]+ 2Tb[a0]+ Ta0b+ Tb0a. (60)

Combining (60) with (49), in which we raise the index and set i = 0, we arrive at

2T(ab)0= 2Ta[b0]+ 2Tb[a0]+ vaTb00+ vbTa00,(61)

which should be compared to (74) in [48]. Remembering that

2T[a0]0= Ta00,(62)

which follows directly from δx0= 0, we can rewrite (61) as follows:

T(ab)0= v(aΛb)0+ ρam∆[0b]m+ ta[0b]+ ρbm∆[0a]m+ tb[0a],(63)

where we made use of (53) and introduced the following definition for the antisymmetric part of the integrated orbital

momentum on the basis of the canonical momentum7

Λab:= 2T[ab]0.(64)

Remembering that tabis a symmetric quantity, equation (63) can be rewritten as

T(ab)0= v(aΛb)0+ ρam∆[0b]m+ ρbm∆[0a]m, (65)

which is analogous to equation (76) in [48]. This result can be used to rewrite (59),

Tb(ai)= Ta[bi]+ Ti[ba]+ vaT[bi]0+ viT[ba]0

+vb?

Ta[i0]+ Ti[a0]+ vaT[i0]0+ viT[a0]0?

, (66)

which resembles the first part of (77) in [48] and can finally be brought into the form

Tb(ai)= ρbm

?

−v(aΛi)m+ ρan∆[im]n+ ρin∆[am]n?

, (67)

which is analogous to the second part of (77) in [48]. The last equation can be used in (53) to obtain

?

After reinsertion into (53) we arrive at the final result,

?1

takl= ρab

∆(kl)b+ v(lΛk)b− ρln∆[kb]n− ρkn∆[lb]n?

.(68)

Talk= ρab

2∆lkb+ ∆klb+ v(lΛk)b− ρln∆[kb]n− ρkn∆[lb]n

?

,(69)

7Note that this definition corresponds to the quantity Labin [48]. In this work, in contrast to [48], we use the symbol Labfor the

“complete” first moment of the integrated canonical momentum, i.e., including also the symmetric part.

Page 13

13

which closely resembles the form of one of the propagation equations found [48], i.e., equation (56). With the help of

(53), (65), and (69), equation (48) can now be transformed into

?1

−Γijkρab

?

where we introduced Pi:= Ti0for the integrated 4-momentum. Equation (70) is analogous to the propagation

equation (55) in [48]. With the help of (70) we can can bring (54) into the form

?1

−Γljcρkb

?

which can be viewed as the analogue to (79) in [48]. Because of the different symmetries in metric-affine gravity

the method used in this section, which was outlined in [48], does not lead to a very compact form of equation (54).

The last equation in the rewritten set is the one relating the time derivative of the momentum to the other matter

quantities; from (47), (52), (54), and (53) we obtain

Ti

a= vaPi+d

dt2Λai+ gil

?1

∆(jk)b+ v(kΛj)b− ρkn∆[jb]n− ρjn∆[kb]n?

?

v(aΛl)0+ ρam∆[0l]m+ ρlm∆[0a]m??

2∆kjb+ ∆jkb+ v(kΛj)b− ρkn∆[jb]n− ρjn∆[kb]n

?

−Nijkρab

, (70)

∇v∆kl0= tkl− vkPl+d

dt2Λkl+ gln

?

v(kΛn)0+ ρkm∆[0n]m+ ρnm∆[0k]m??

?1

2∆cjb+ ∆jcb+ v(cΛj)b− ρcn∆[jb]n− ρjn∆[cb]n

∆(jc)b+ v(cΛj)b− ρcn∆[jb]n− ρjn∆[cb]n?

?

−Nljcρkb

, (71)

d

dtPi= Rijkl?

vj∆kl0+ ρjn∆kln?

+Γijk?∇v∆jk0+ ρnl

+Γijk,aρab∆jkb+

?Γnmj∆mkl− Γnkm∆jml??

Γ ijktjk+

Γ ijk,atajk.

{ }

{ }

(72)

This equation should be compared to (80) in [48]. We only note that an elimination of tjk and tajkin the last two

terms of (72) is possible by using (53) and (70). In the next section we work with a slightly different set of quantities,

which allow for a very condensed form of the propagation equations of metric-affine gravity.

D.Rewriting the propagation equations

In this section we present a more condensed form of the propagation equations. Thereby we find a direct general-

ization of the main result8of [48], i.e., equations (53)-(58), in the case of metric-affine gravity.

We introduce the following notation for the integrated quantities: Pi:= Ti0denotes again the integrated 4-

momentum, Lkl := Tkl0the total orbital canonical energy-momentum and Ykl := ∆kl0the integrated intrinsic

hypermomentum. Furthermore, recalling that the hypermomentum comprises the spin, dilaton charge, and intrinsic

shear, it is convenient to denote the antisymmetric part of the hypermomentum as the integrated spin τkl:= ∆[kl]0,

whereas the trace of the hypermomentum defines the integrated dilaton charge Z := ∆kk0.

In addition, we introduce a shorter notation for the “convective currents,” i.e., the projected quantities which we

have used in previous sections and which are also used in [48]. For the intrinsic hypermomentum, we have

(c)

∆klm:= ∆klm− vm∆kl0≡ ρmn∆kln,

and for the orbital canonical energy-momentum

(c)

Tklm:= Tklm− vmTkl0≡ ρmnTkln.

8Please note the typo in equation (53) of [48]. Using the notation of [48] the last term in (53) should read: ... +1

2ρδνNβανgγǫ∇ǫλαβδ.

Page 14

14

The convective spin and dilaton currents arise as the antisymmetric part and the trace of the convective current of

the intrinsic hypermomentum, i.e., as

(c)

τklm:= ∆[kl]m− vmτkl

and

(c)

Zk:= Zk− vkZ,

respectively (here Zk:= ∆jjk). With this notation, we recast the propagation equations (48)-(51) into

Tk

i= viPk+d

dtLik−

{ }

Γ kjlTilj+ Nkjl(c)

∆jli, (73)

(c)

T(aib)= 0,(74)

∇vYik = −Tk

(c)

∆kla= Talk− takl.

i+ tik− Γjli(c)

∆lkj+ Γjkl(c)

∆ilj,(75)

(76)

Equation (73) describes the canonical energy-momentum in terms of the usual combination of the “translational”

plus “orbital” contributions (the first two terms), plus the additional contribution of the first moments. One should

compare this with the alternative formula (70). Equation (74) simply tells us that the convective current

antisymmetric in the upper indices a and b. This is a useful technical fact. The next equation (75) is actually an

equation of motion for the intrinsic hypermomentum. Its form closely follows the Noether conservation law of the

hypermomentum, cf. (19) and (21). An alternative form of such a dynamical equation for the hypermomentum is

given in (71). Finally, the equation (76) expresses the convective intrinsic hypermomentum current in terms of the

first moments of the energy-momentum.

Equations (73)-(76) are easily derived from (48)-(51), one only needs to rearrange some terms. In contrast to this,

we need some additional steps to arrive at a new form of equation (47), which represents the most interesting of the

propagation equations from a physical point of view.

We start by expanding the general connection in (47), this yields

(c)

Taibis

d

dtTi

0= Rijkl∆klj+

{}

Γ iklTl

k− Nikl?Tl

k− tkl

?+

{}

Γ ikl,aTalk− Nikl,a

?Talk− takl

?.(77)

Furthermore, we have

d

dt

?

Ti

0− Nikl∆kl0?

=d

dtTi

0− vaNikl,a∆kl0− Nikld

dt∆kl0.(78)

Insertion of (50) and (77) into (78) yields

d

dt

?

Ti

0− Nikl∆kl0?

= Rijkl∆klj+

−Γmjk∆jlm?− Nikl,a

=

Γ iklTl

{ }

Γ iklTl

k+

{ }

Γ ikl,aTalk− Nikl?

?

Γjlm∆kmj

Talk− takl+ va∆kl0?

(79)

{}

k+

{}

Γ ikl,aTalk+ ∆klj?Rijkl− ΓjplNikp+ ΓjkpNipl

Γ jipNpkl−−Nikl,j+

{}

{}

Γ jipNpkl

?

.(80)

In the last step we made use of (51) in order to replace the terms in the last brace in the second line of (79).

Furthermore, we added a “0” dummy term, i.e., the last two terms in the second line of (80). We proceed by replacing

the curvature by its decomposition, i.e.,

Rijkl=

{ }

Rijkl+

{}

∇jNikl−

{}

∇iNjkl+ NiplNjkp− NjplNikp,(81)

Page 15

15

equation (80) then turns into

d

dt

?

Ti

0− Nikl∆kl0?

=

{}

Γ iklTl

k+

{ }

Γ ikl,jTjlk+ ∆klj

?{ }

Rijkl−

{}

Γ jipNpkl−

{}

∇iNjkl

?

.(82)

We rewrite (48) with the help of (51):

Tl

k=

d

dtTkl0+ vkTl

0−

{ }

Γ lpmTkmp+ Nlpm?∆pmk− vk∆pm0?.(83)

Contracting this equation with the Levi-Civita connection and introducing another “0” dummy term yields

?{}

−va{}

+

Γ lpm−

{}

Γ iklTl

k=

d

dt

Γ iklTkl0

?

+

{}

Γ iklvk

?

Tl

0− Nlpm∆pm0−

{ }

Γ lpmTpm0

?

Γ ikl,aTkl0−

?{ }

{}

Γ ikl{}

Γ ipl{}

Γ lpm

(c)

T

kmp+

{}

Γ iklNlpm∆pmk

Γ ikl{ }

{}

Γ lkm

?

Tpm0vk. (84)

With this result at hand we can replace the first term on the rhs of (82), i.e.,

?

?{ }

If we introduce a new quantity

d

dt

Ti

0− Nikl∆kl0−

{}

Γ iklTkl0

?(c)

?

−

{}

Γ iklvk

?{ }

?

Tl

0− Nlpm∆pm0−

?

{ }

Γ lpmTpm0

?

=

Γ ikl,j−

{}

Γ ijp{}

Γ pkl

T

jlk+ ∆klj

Rijkl−

{ }

∇iNjkl

+

{ }

RkjilTkl0vj. (85)

Pi:= Ti

0− Nikl∆kl0−

{}

Γ iklTkl0

(86)

as a generalized total 4-momentum, equation (85) can be written in a more compact form as follows:

{}

∇vPi=

{ }

RkjilTkl0vj+

?{}

Γ ikl,j−

{ }

Γ ijp{}

Γ pkl

?(c)

T

jlk+ ∆klj

?{}

Rijkl−

{}

∇iNjkl

?

. (87)

By using the Ricci identity

{}

Rjkil+

{}

Rkijl+

{}

Rijkl= 0 and the fact that the convective part of first integrated moment

(c)

T

?{ }

of the canonical-momentum is antisymmetric in the upper two indices, i.e.,

kmp=

(c)

T

[kmp], we can recast (87) into

{}

∇vPi=

{}

RkjilTkl0vj+

Rijkl−

{ }

∇iNjkl

?

∆klj+

{ }

Rijkl

(c)

T

klj.(88)

This equation represents the rewritten form of (47) and should be compared to (72) in the previous section.

It is worthwhile to notice the general feature that characterizes the coupling between the physical objects (currents)

with the geometrical objects (metric, connection, and the derived quantities). Namely, the intrinsic current (the one

that is truly microscopic, which arises from the averaging over the medium with the elements with microstructure, i.e.,

that possess internal degrees of freedom) couples to the post-Riemannian geometric quantities, see the second term

on the rhs of (86) and the first term on the rhs of (88). In contrast to this, the orbital canonical energy-momentum

(which is induced by the macroscopic dynamics of the rotating and deformable body) is only coupled to the purely

Riemannian geometric variables and never couples to the post-Riemannian geometry, see the last terms on the right-

hand sides of (86) and (88). This observation represents a generalization of the result of Yasskin and Stoeger [48], in

other words, it proves that the possible presence of the post-Riemannian geometry (in particular, of the torsion and

the nonmetricity) can only be tested with the help of the bodies that are constructed from media with microstructure

(spin, dilaton charge, and intrinsic shear). Test particles composed from usual matter, i.e., without microstructure,

are not affected by the post-Riemannian geometry, and they thus cannot be used for the detection of the torsion and

the nonmetricity.

In order to get a better understanding of this fact, we will consider several special cases of the metric-affine geometry

in the subsequent sections, moving from a general non-Riemannian geometry back to the Riemannian one.

Page 16

16

VI.RELATION BETWEEN THE INTEGRATED MOMENTS

In different situations, it is technically convenient to use different definitions of the integrated moments (see also

[85] for the behaviour under infinitesimal coordinate transformations). However, directly from the definitions (32)

and (43) we can establish relations between two sets of the moments.

Starting with the identity˜tij= gjk˜tik, we expand the metric in the same way as the other geometric quantities

(46),

gjk??

and then by integration over the world tube, in the pole-dipole approximation we find

x= gjk??

Y+ δxagjk,a

??

Y+ ···,(89)

tij= tij− 2

{}

Γ l(kj)tlik. (90)

We used here the metricity condition gjk,l= −

Analogously, we have for the integrated canonical energy-momentum

{}

Γ lnjgnk−

{}

Γ lnkgjn.

T

ij= Tij− 2

{}

Γ l(ik)Tlkj. (91)

The “inverse” formulas read

tij = ti

j+ 2

{}

Γl(jk)tlik,(92)

Ti

j= Tij+ 2

{}

Γl(ik)T

lkj. (93)

Hence, in the pole-dipole approximation, the integrated hypermomenta and the first moments of the canonical and

metrical energy-momenta in both sets are the same:

∆ijk= ∆

tijk= tijk,

ijk, (94)

(95)

Tijk= T

ijk.(96)

With the help of (95) and (96), we can verify the consistency of the relations (90) and (92), as well as (91) and (93).

For single-pole test particles, the corresponding integrated energy-momenta coincide since the last terms in (90)-(93)

vanish.

VII.CONCLUSIONS & OUTLOOK

In this work we derived the equations of motion for test particles in metric-affine gravity from the conservation

laws of the theory with the help of a multipole formalism. Apart from the general form of the equations of motion,

we explicitly presented the propagation equations for pole-dipole test particles. Our results are valid for a very large

class of gravitational theories, i.e., all theories which fit into the framework of metric-affine gravity. The equations

derived in this work should be used to systematically study the motion of test particles with spin, shear, dilation,

and rotation within alternative gravitational theories in a non-Riemannian context. Our results generalize previous

analyses [48, 86, 87, 88, 89], which were carried out in the context of general relativity, Einstein-Cartan theory, and

within Poincar´ e gauge theory.

A. Special cases

In this section we discuss several special cases within our framework by either making assumptions about the internal

structure of the test particles, or by constraining the background geometry. The full agreement, in some special cases,

with the well-known results from general relativity and Poincar´ e gauge theory demonstrates the consistency of our

framework.

Page 17

17

1.Equations for a single-pole particle in metric-affine gravity

Let us consider the propagation equations for a single-pole test particle in metric-affine gravity, i.e., the set (36)-(40)

with vanishing dipole contributions:

d

dtT

vaT

d

dt∆kl0

va∆

i0

= Rijkl∆

klj− NiklT

lk−

{}

Γ kjiT

(kj)+ Nikltkl,(97)

i0

= T

ia,(98)

= Nmjk∆jlm−

kla.

{}

Γ mjk∆mlj−

{}

Γ mjl∆k(mj)− Njlm∆k

mj− T

lk+ tkl, (99)

kl0

= ∆ (100)

It is a common folklore that in generalized gravity theories the equation of motion for single-pole test particles is given

by some kind of “generalized” geodesic equation. By generalized we mean an equation which has the same form as

the geodesic equation, i.e., the equation of motion for single-pole test particles in general relativity, but in which the

Levi-Civita connection has been replaced by the full (non-Riemannian) connection. The result in (97)-(100) clearly

demonstrates that such an assumption is not substantiated.

a. Particles without intrinsic hypermomentum

If we perform a further specialization by considering only test

particles without intrinsic hypermomentum, the set (97)-(100) turns into

d

dtT

vaT

i0

= −NiklT

= T

= tkl.

lk−

{}

Γ kjiT

(kj)+ Nikltkl,(101)

i0

ia,(102)

T

lk

(103)

Of course the first and the last term on the rhs of (101) cancel because of (103) and the equation of motion for a

test particle without intrinsic hypermomentum is then given by the regular geodesic equation [in the next section we

explicitly show how one can recover the geodesic equation from the set (101)-(103)]. This generalizes the well-known

result from Poincar´ e gauge theory to metric-affine gravity, i.e., a test particle without intrinsic hypermomentum will

not “feel” the torsion or the nonmetricity of the underlying spacetime. Hence, test particles without intrinsic spin,

shear, or dilation current are not suitable for mapping the non-Riemannian features of spacetime. Accordingly, current

experiments like Gravity Probe-B [90] are not suitable for the detection of torsion in contrast to what is sometimes

claimed by other authors. At this point, one should mention that a coupling between torsion and matter without

intrinsic spin currents may be achieved in some nonstandard gravity theory, although the authors of the present

paper are not aware of any viable candidate for such a theory. For any theory which fits into the very general and

well-motivated framework of metric-affine gravity, e.g., Poincar´ e gauge theory and Einstein-Cartan theory, such a

coupling will not occur.

2.Recovering the geodesic equation

In this section we explicitly show that the single-pole equations of motion for a test particle without intrinsic

hypermomentum take the form of the usual geodesic equation. The set (101)-(103) reduces to

d

dtT

i0

= −

= vaT

= tkl.

{}

Γ kjiT

(kj), (104)

T

iai0, (105)

T

lk

(106)

Now lets us introduce the velocity ua:= dYa/ds along the world line of the particle.

ds2= gabdYadYb, and remember that Ya(t) = xa(Y (t)) = tδa

definition we can rewrite (104) and (105) as follows:

Note that u0= dt/ds,

0, d/dt = va∂a, uaua= 1, va= dYa/dt. With this

d

dsT

i0+

{}

Γ kjiu0T

kj

= 0,(107)

u0T

ia

= uaT

i0.(108)

Page 18

18

Setting i = 0 in the last equation and reinsertion into (107), together with the definition m := T

ia= muiua. This in turn can be used to rewrite (107) as follows:

00/?u0?2, yields

T

d

ds

?mui?+

{}

Γ kjimukuj= 0.(109)

Multiplication of this equation by ui and remembering that ub{ }

∇bua=

?

ua,b+

{ }

Γ cbauc

?

ub, dua/ds = ua,bub,

ua

{}

∇bua= 0 yields

dm

dsuiui+ mukuj{}

∇juk= 0=⇒

dm

ds

= 0.(110)

When we use this result in (109) we end up with

dui

ds+

{}

Γ kjiukuj= 0,(111)

which is the geodesic equations. Hence, in metric-affine gravity single-pole test particles without intrinsic hypermo-

mentum, i.e., without spin, shear, and dilation currents, move in exactly the same way as test particles in general

relativity. We stress that no constraining assumptions about the geometry of the background spacetime have been

made in order to derive this result. Equation (111) is valid in a completely general metric-affine spacetime, i.e., the

background can be a non-Riemannian one with nonvanishing torsion and nonmetricity, the test particle just does not

feel this geometric features as long as it does not posses any “microstructure” in the form of a nonvanishing intrinsic

hypermomentum.

In later sections we are also going to discuss the equations of motion for some special cases in which we impose an

a priori restriction on the geometry of the background spacetime.

3.Recovering the Mathisson-Papapetrou equations

Also the well-known propagation equations for a classical pole-dipole test particle can be easily recovered in our

framework. For particles without intrinsic hypermomentum in a Riemannian background the propagation equations

in (36)-(40) turn into

d

dtT

d

dtT

i0

= −

{}

Γ kjiT

(kj)−

{}

Γ kji,aT

a(kj), (112)

ai0

= T

ia− vaT

bia+ T

i0−

aib,

{ }

Γ kjiT

a(kj),(113)

vaT

bi0+ vbT

ai0

= T

= tkl,

= takl.

(114)

T

lk

(115)

T

alk

(116)

These equations are exactly the equations of motion for a pole-dipole particle described by Papapetrou in (3.2)-(3.4)

of [21]. This result clearly demonstrates the consistency and generality of our framework.

4. Propagation equations in a Weyl-Cartan spacetime

The Weyl-Cartan spacetime is characterized by a special type of nonmetricity, namely, when the 1-form of the

nonmetricity Qαβ= gαβQ reduces to just the Weyl covector Q = Qidxi. Correspondingly, the distorsion 1-form then

reduces to

Nαβ= −1

2δβ

αQ + Kαβ,(117)

where the contortion Kαβ = −Kβα:= N[αβ]is just the antisymmetric piece of the distorsion (note, however, that

Kαβis constructed from both the torsion and the Weyl nonmetricity). In components, we have explicitly Niαβ=

−1

2δβ

αQi+ Kiαβ.

Page 19

19

Using relation (117), we derive the propagation equations for test particles on the background of the Weyl-Cartan

spacetime:

?{ }

i= viPk+d

dtLik−

(c)

T(aib)= 0,

Γ jli(c)

∆lkj+

{}

∇vPi =

{ }

RkjilTkl0vj+

Rijk

l−

{}

∇iKjkl

?

τjli−1

τklj+

{ }

Rijk

l(c)

Tklj+1

2(

{}

∇iQj)Zj,(118)

Tk

{ }

Γ kjlTilj+ Kkjl(c)

2Qk

(c)

Zi, (119)

(120)

{}

∇vYik = −Tk

(c)

∆kla= Talk− takl.

i+ tik−

{ } {}

Γ jkl(c)

∆ilj+ Kjli∆lkj− Kjkl∆ilj, (121)

(122)

Here Pi= Pi+1

a.Single-pole particles

surprisingly simple system

2QiZ − Kiklτkl−

{}

Γ iklLkl.

For the single-pole case (when all of the first integrated moments vanish), we find a

{ }

∇vPi+ KijkvjPk= Rijklvjτkl+1

Tk

∇vYik = −Tk

(c)

∆kla= 0.

2fijvjZ −1

2QidZ

dt,

(123)

i= viPk,(124)

(125)

i+ tik,

(126)

Here we introduced fij:= ∂iQj−∂jQi. Thus, provided a test particle has a nontrivial integrated dilaton charge Z, it

will be affected in the Weyl-Cartan spacetime by the Lorentz–type force represented by the second term on the rhs of

the propagation equation (123). If, in addition, the test particle has a nontrivial spin τkl, the latter will be affected

by the Mathisson-Papapatrou–type force which is determined by the Weyl-Cartan curvature, as described by the first

term on the rhs (123).

5.Propagation equations in a Weyl spacetime

Weyl [2, 91, 92] was the first who noticed a similarity between the electromagnetic vector potential and the non-

metricity covector Qi. Indeed, this is also manifested in the equations of motion, as becomes clear from the rhs of

equation (123). However, an essential difference is that the Weyl nonmetricity may interact with the dilaton charge

and not with the electromagnetic charge.

The Weyl geometry arises as a special case of the Weyl-Cartan spacetime, when the torsion Sijk:= Γijk−Γjik= 0

is equal zero.9In this case the distorsion is still given by (117), but the contortion is expressed in terms of the Weyl

covector only:

?gijQk− δk

The propagation equations in the Weyl spacetime are formally the same as (118)-(122) where we have to substitute

the contortion (127). Analogously, the dynamics of single-pole test particles is described in the Weyl spacetime by

(123)-(125) with (127) inserted.

Kijk=1

2

iQj

?. (127)

6.Propagation equations in a Riemann-Cartan spacetime

The Riemann-Cartan spacetime arises from the Weyl-Cartan geometry for the case of vanishing nonmetricity,

Qi= 0. The distorsion then coincides with the contortion and is constructed only from the torsion: Nijk= Kijk=

1

2(Sjki+ Sikj+ Sjik).

9Our notation for the spacetime torsion is different from [50]. Since we reserved the symbol T for energy-momentum related objects, the

torsion tensor is here denoted by the symbol S as in the old review [93].

Page 20

20

The propagation equations for pole-dipole particles in Riemann-Cartan spacetime are easily derived by putting

Qi= 0 in equations (118)-(122). We will not write these equations explicitly.

a.Single-pole particles

In order to discuss the propagation equations for single-pole particles, we again introduce

the 4-velocity ua:= dYa/ds along the world line of the particle. With u0= dt/ds and ds2= gabdYadYb, we have

uaua= 1 (note that ua= u0va). Then, it is straightforward to verify that in the Riemann-Cartan spacetime equations

(123)-(125) reduce to

˙Pi = SijkujPk+ Rijklujτkl,

u0Tk

˙ τij = u[iPj],

˙Y(ij) = u0?

(128)

i= uiPk,(129)

(130)

t(ij)− T(ij)

?

. (131)

Here we denoted the covariant (Riemann-Cartan) derivative along the trajectory by a dot: “˙”= D/ds = ui∇i.

It is satisfactory to see that with (128) and (130) we recover the usual equations of motion for a test particle

with mass and spin in the Riemann-Cartan spacetime [86, 87, 93]. One should note, however, that we are still

in the framework of the metric-affine gravity in which a test particle carries, besides the mass and spin, also the

dilaton charge and the intrinsic shear. The latter integrated characteristics are described by the symmetric part

of the intrinsic hypermomentum Y(ij). The dynamics of these quantities is determined by equation (131) which is

completely decoupled from the other propagation equations. In other words, they do not affect the motion of a test

particle in the Riemann-Cartan spacetime, and the trajectory is completely defined by the behavior of the integrated

4-momentum Piand the integrated spin τkl.

Let us contract equation (130) with ui. This then yields the explicit form of the integrated 4-momentum,

Pj= muj+ 2ui˙ τij,(132)

where we introduced the notation for the rest mass of the body m := uiPi(i.e., the momentum projected to the rest

frame). By substituting this back into (130) we obtain the dynamical equation for the spin

˙ τij− uiuk˙ τkj+ ujuk˙ τki= 0.(133)

7. Propagation equations in a Riemannian spacetime

When all the post-Riemannian geometric objects are trivial (no torsion and no nonmetricity, i.e., Nαβ= 0), the

propagation equations on the purely Riemannian spacetime reduce to

{ }

∇vPi =

{}

RkjilTkl0vj+

{}

Rijk

l?

τklj+

(c)

Tklj?

, (134)

Tk

i= viPk+d

dtLik−

{}

Γ kjlTilj, (135)

(c)

T(aib)= 0, (136)

{}

∇vYik = −Tk

(c)

∆kla= Talk− takl.

i+ tik−

{}

Γ jli(c)

∆lkj+

{}

Γ jkl(c)

∆ilj, (137)

(138)

Here Pi= Pi−

a.Single-pole particles

{ }

Γ iklLkl.

For the single-pole particles with vanishing intrinsic hypermomentum this simplifies to

{}

∇vPi= 0,

Tk

Tk

(139)

(140)

(141)

i= viPk,

i= tik.

The resulting trajectories are geodesics.

Page 21

21

8.Propagation equations in a Riemannian spacetime (alternative form)

For completeness let us also determine the explicit form of the propagation equations using the upper-index con-

vention for the integrated moments, for the special case of a Riemannian background. From (36)-(40) we can infer

that pole-dipole particles move according to

d

dtT

vaT

d

dt∆

i0

=

{}

R

ijkl∆

klj−

{}

Γ kjiT

(kj),(142)

i0

= T

ia,(143)

kl0

= −

{}

Γ mjk∆

mlj−

kl0− T

{}

Γ mjl∆

k(mj)− T

alk+ takl.

lk+ tkl,(144)

0 = ∆

kla− va∆(145)

a.Single-pole particles

brings the set (142)-(145) into the form

Further restriction to single-pole particles with vanishing intrinsic hypermomentum ∆abc

d

dtT

i0

= −

= vaT

= tkl.

{}

Γ kjiT

(kj),

T

iai0,

T

lk

As we have already shown these equations lead to the geodesic equation. Hence, within our general formalism we can

quickly reproduce the standard result of general relativity.

B.Open problems

The results obtained in this work are valid for a wide class of extended gravitational theories that are naturally

embedded into the framework of metric-affine gravity. However, our study is not exhaustive in many important

aspects, and at this stage there remain several interesting open questions related to the multipole expansion of the

equations of motion of test particles in alternative gravity theories.

1.Invariant definition of moments

As we have already mentioned in previous sections, the definition of the integrated moments of the matter currents

in the multipole formalism is to a certain extent ambiguous. This is related to the index positions in the integrand

expression and to the nonconstancy of the metric which is used to lower and raise the indices. In view of this problem,

we decided to present the full set of propagation equations for two different choices of the integrated moments, defined

in equation (32) and (43), respectively. Thereby one covers the definitions which have been discussed most frequently in

the literature. Although we clearly favor the definition with mixed indices (43), for the formal reasons given in section

V, even other index positions than the ones investigated in the present work are imaginable. Such an ambiguity

in the definition of the integrated moments motivates the search for an invariant formulation. The corresponding

program was already carried out in several works within a general relativistic context [31, 36, 38, 40, 94]. Within

an alternative gravity theory like metric-affine gravity, which is no longer a purely metric theory but has a richer

geometrical structure, a detailed investigation is needed in order to generalize the concepts linked to such an invariant

formulation.

2. Supplementary conditions

Previous analyses [31, 95, 96, 97] in metric theories of gravitation have shown, that even at the dipole level,

supplementary conditions are needed in order to obtain a closed set of propagation equations. Indeed, let us recall

the propagation equations in the Riemann-Cartan spacetime, for example. The four equations (132) are sufficient

to find the four coordinates of a position of a particle on its trajectory. However, the system (133) contains only

three independent equations, and this is not sufficient to determine six components of the spin. As a result, the

Page 22

22

supplementary conditions are usually imposed on the spin of the test particles in order to make number of the equations

equal to the number of unknown variables. The imposition of an additional supplementary condition comes with some

assumptions about the physical nature of the particles under consideration, and there is no unique prescription how

to do it. Even within the context of general relativity, a number of competing conditions exists. Furthermore, there

seems to be no consensus on which of the supplementary conditions is the most physical one. In the context of

alternative gravity theories the spectrum of possible supplementary conditions is greatly enhanced. This fact can

be ascribed to the additional degrees of freedom within such theories, in particular, regarding the matter variables

describing the internal structure of particles. Although there exist several studies of such supplementary conditions in

the literature, most of them in the context of Einstein-Cartan and Poincar´ e gauge theory, a systematic and up-to-date

analysis in the context of metric-affine gravity is still an outstanding task. We only note that an ultimate judgment

over the correct choice of a supplementary condition can only be made with the help of an experiment.

3. Propagation equations involving higher moments

If we take into account previous results in Einstein’s theory [98], it is to be expected that the role of supplementary

conditions is even aggravated at higher orders of approximation. Of course this is due to the fact that at higher orders

we need an even more detailed description of the internal dynamics of the test particles. Nevertheless, the study of

higher orders of the propagation equations, beyond the pole-dipole level, will be of great interest in the context of

radiation phenomena. In particular, we expect that such studies will shed light on our understanding of the new field

strengths of metric-affine gravity, i.e., torsion and nonmetricity, which have no counterpart in the classical theory

gravitation, namely general relativity.

4.Relation to other approximation schemes

From a more formal standpoint, we can also ask about the compatibility with other approximation schemes which

were employed in the context of gravitational theories. The most prominent examples being the post-Minkowskian and

post-Newtonian approximation. Since these approximation schemes, in their full generality, are still under construction

in the context of metric-affine gravity, a systematic comparison with the results obtained within a multipole scheme

appears to be a long term project.

To sum up, the study of the propagation equations of deformable test particles with the help of a multipole

approximation scheme is a very rich field of research. In the context of alternative gravity theories this field is still

in its infancy. Apart from the first steps undertaken in this work a number of open problems remain; we intend to

attack these in future works.

Acknowledgments

The authors are grateful to F.W. Hehl (Univ. Cologne) for stimulating discussions and constructive criticism. Y.N.O.

was supported by the Deutsche Forschungsgemeinschaft (Bonn) with the grant HE 528/21-1. D.P. acknowledges the

support by Ø. Elgarøy (Univ. Oslo) and the Research Council of Norway under the project number 162830.

APPENDIX A: GENERAL CONVENTIONS AND NOTATIONS

In the theory of metric-affine gravity, the gravitational field is described by the three basic variables: the metric

gαβ, the coframe ϑα, and the linear connection Γαβ. The Latin indices i,j,... are used for local holonomic spacetime

coordinates and the Greek indices α,β,... label (co)frame components. The vector basis dual to the frame 1-forms

ϑαis denoted by eαand they satisfy eα⌋ϑβ= δβ

an exterior form. Using local coordinates xi, we have ϑα= hα

carry the local Lorentz indices can be recast into their counterparts with the coordinate indices with the help of the

contraction with the components of the tetrads, hα

α.

α. Here ⌋ denotes the interior product (contraction) of a vector with

idxiand eα= hi

α∂i. All objects and equations that

iand hi

Page 23

23

1.Geometrical objects

The geometry of MAG is described by the curvature 2-form Rαβ:= dΓαβ+ Γγβ∧ Γαγ, the nonmetricity 1-form

Qαβ:= −Dgαβ, and the torsion 2-form Tα:= Dϑαwhich are the gravitational field strengths for linear connection

Γαβ, metric gαβ, and coframe ϑα, respectively.

It is convenient to define a 1-form tensor-valued difference of the Riemannian (Christoffel) connection and the

general linear connection:

Nαβ:=

{ }

Γ αβ− Γαβ.(A1)

This quantity is known as distorsion 1-form. In particular, the torsion is recovered from it as Tα= −Nβα∧ ϑβ,

whereas the nonmetricity arises as Qαβ= −2N(αβ). The corresponding curvature 2-forms are related via

Rαβ=

{}

Rαβ−

{}

DNαβ+ Nγβ∧ Nαγ.(A2)

2.Physical objects

The sources of the metric-affine gravitational field are the 3-forms of the canonical energy-momentum and hyper-

momentum. They are defined by the variational derivatives of the material Lagrangian 4-form Lmat, respectively:

Σα =

δLmat

δϑα,

δLmat

δΓαβ.

(A3)

∆αβ =

(A4)

The Lagrangian Lmatalso depends on some matter fields ψ, but this is irrelevant for the current discussion.

3. Components

When the local coordinates xiare chosen, we can write all the geometrical and physical quantities explicitly in

terms of their components:

ϑα= hα

Γαβ= Γiαβdxi,

Nαβ= Niαβdxi,

idxi, (A5)

(A6)

(A7)

Rαβ=

1

2Rijαβdxi∧ dxj,

Σα = Tαi∂i⌋η,

∆αβ = ∆αβi∂i⌋η.

(A8)

(A9)

(A10)

Here η is the volume 4-form. Writing the Lagrangian form as Lmat= Lmatdx0∧ dx1∧ dx2∧ dx3, we can recast the

definitions (A3) and (A4) as follows:

?Tαi=

?∆αβi=

δLmat

δhα

δLmat

δΓiαβ.

i

,(A11)

(A12)

APPENDIX B: DIMENSIONS & SYMBOLS

In order to fix our notation, we provide some tables with definitions in this appendix. The dimensions of the

different quantities appearing throughout the work are displayed in table II. Table III contains a list with symbols

used throughout the text.

Page 24

24

TABLE II: Dimensions of the quantities within this work.

Dimension (SI)Symbol

Geometrical quantities

1

m

m−1

m−2

m4

gαβ, δαβ, gij,√−g, hα

xi, dxi, ds, δxi, Ya, ϑα, Tα

eα, Γiαβ, Niαβ, Kiαβ, Qiαβ, Qi, Sijk

Rijαβ, fij

η

i, Γαβ, Nαβ, Kαβ, Rαβ, Qαβ, Q, ℓαβ

Matter quantities

1

kgm2/s

uα, va, ρab, ψ

h (Planck constant), L, Lmat, Ltot, ∆αβ, ταβ, σαβ, mαβ, Mαβ, Hαβ, Eαβ, ∆ijk, ∆

T

Hα, Eα, Σα, Ti

∆αβi

Tαi, Lmat

ijk, Tijk, tijk,

ijk, tijk, Ykl, τkl, Lkl, Λkl, τklj, Zk, Z

k, tij, Tkgm/s

kg/(m s)

kg/(m2s)

ij, tij, Pi, Pi, m

Operators

1d, D

m−1

∂i, ∇i, ∇v,

{ }

? Lξ

[1] A. Einstein and J. Grommer. Allgemeine Relativit¨ atstheorie und Bewegungsgesetz. Sitzungsb. Preuss. Akad. Wiss., page 2,

1927.

[2] H. Weyl. Raum-Zeit-Materie. Springer-Verlag, Berlin, 1923.

[3] A. S. Eddington. The mathematical theory of relativity. Cambridge University Press, London, 1924.

[4] A. E. Scheidegger. Gravitational motion. Rev. Mod. Phys., 25:451, 1953.

[5] J. N. Goldberg. The equations of motion. Gravitation: An introduction to current research, edited by L. Witten, Wiley,

New York, page 102, 1962.

[6] P. Havas. The early history of the “problem of motion” in General Relativity. Einstein Studies, 1:234, 1986.

[7] T. Damour. The problem of motion in Newtonian and Einsteinian gravity. 300 Years of Gravitation, Cambridge University

Press, edited by S.W. Hawking and W. Israel, page 128, 1987.

[8] A. Einstein, L. Infeld, and B. Hoffmann. The gravitational equations and the problem of motion. Ann. Math., 39:65, 1938.

[9] A. Einstein, L. Infeld, and B. Hoffmann. Appendices to ‘The gravitational equations and the problem of motion’. Hand-

written supplement (IAS Library), 1938.

[10] V. A. Fock. Sur le mouvement des masses finies d’apr` es la th´ eorie de gravitation einsteinienne. J. Phys. (Moscow), 1:81,

1939.

[11] M. Mathisson. Neue Mechanik materieller Systeme. Acta Phys. Pol., 6:163, 1937.

[12] D. Puetzfeld. The cosmological post-Newtonian equations of hydrodynamics in General Relativity. Unpublished, 2007.

[13] C. Lanczos. Zur Dynamik der allgemeinen Relativit¨ atstheorie. Z. Phys., 44:773, 1927.

[14] M. Mathisson. Die Beharrungsgesetze in der allgemeinen Relativit¨ atstheorie. Z. Phys., 67:270, 1931.

[15] M. Mathisson. Die Mechanik des Materieteilchens in der allgemeinen Relativit¨ atstheorie. Z. Phys., 67:826, 1931.

[16] M. Mathisson. Bewegungsproblem der Feldphysik und Elektronenkonstanten. Z. Phys., 69:389, 1931.

[17] H. P. Robertson. Test corpuscles in general relativity. Proc. Edn. Math. Soc., 5:63, 1937.

[18] A. Papapetrou. Gravitationswirkungen zwischen Pol-Dipol Teilchen. Z. Phys., 116:298, 1940.

[19] C. Lanczos. The dynamics of a particle in General Relativity. Phys. Rev., 59:813, 1941.

[20] L. Infeld and A. Schild. On the motion of test particles in General Relativity. Rev. Mod. Phys., 21:408, 1949.

[21] A. Papapetrou. Spinning test-particles in General Relativity. I. Proc. Roy. Soc. London Ser. A: Math. Phys. Sci., 209:248,

1951.

[22] A. Papapetrou and W. Urich. Das Pol-Dipol-Teilchen im Gravitationsfeld und elektromagnetischen Feld. Z. Naturforsch.,

10A:109, 1955.

[23] A. Papapetrou. Equations of motion in General Relativity. Proc. Phys. Soc. London A, 64:57, 1951.

[24] A. Papapetrou. Das Problem der Bewegung in der allgemeinen Relativit¨ atstheorie. Fortschr. Phys., 1:29, 1953.

Page 25

25

TABLE III: Directory of symbols.

SymbolExplanation Form degree

Differential form Component

notation

Geometrical quantities

gαβ

gab

g

Metric

Determinant of the metric

Volume form

Coframe

Torsion

Vector basis

Nonmetricity (Weyl 1-form denoted by Q = Qidxi)

0

0

4

1

2

0

1

η

ϑα

Tα

eα

Qαβ

Sijk

Qijk

Rαβ,

{}

Rα

β

Rijkl,

? Rijkl

Nijk

Kijk

Ya

ua

{ }

Rijk

lGeneral curvature, Riemannian curvature

Curvature “object” [defined in eq. (18)]

2

0

Γαβ,

Nαβ

Kαβ

{ }

Γ αβ

Γijk,

{ }

Γ ij

k

Linear connection, Riemannian (Christoffel) connection

Distorsion

Contortion (antisymmetric part of the distorsion)

Worldline within the worldtube of the test particle

Velocity along the worldline Yaof the particle

1

1

1

0

0

Matter quantities

Ltot,L,Lmat

σαβ

Σα

∆αβ

Total, gravitational, matter Lagrangian

Symmetric energy-momentum current

Canonical energy-momentum current

Hypermomentum current

n-th integrated moment of the hypermomentum

n-th integrated moment of the canonical energy-mom.

n-th integrated moment of the symmetric energy-mom.

Generalized integrated momentum

Generalized integrated orbital momentum

Antisymmetric part of the gen. int. orbital momentum

Generalized integrated hypermomentum

Dilaton part, i.e. the trace, of the generalized int. hypermomentum

Spin current (antisymmetric part of the hypermomentum current)

Placeholder for the density of a matter current (e.g.? ∆klj,? Tij, or? tkl)

Generalized total 4-momentum [defined in eq. (86)]

4

4

3

3

0

0

0

0

0

0

0

0

3

0

0

0

tij

Tij

∆ijk

∆

T

tb1···bnij

Pi

L

Λ

Y

Z, Z

τijk

JA

ψ

Pi

b1···bnijk

b1···bnij

ab

ab

ab

k

ταβ

Placeholder for a general matter field

Operators

{ }

D

D,

∇i,

∇v

,i

{ }

∇i

Covariant (exterior) derivative, Riemannian covariant (exterior) derivative

Convective covariant derivative (see, e.g., eq. (55))

Exterior/partial derivative

n → n + 1

n → n + 1

n → n + 1

n → n

0

d

{ }

? Lξ

Riemannian covariant Lie derivative

Spatial projector (equals the convective part, denoted by(c))ρab

Accents

“(c)”

“?”

“

Denotes the convective part of an object

Denotes the density of an object

Denotes integrated version of a density based on upper-index convention

Denotes integrated version of a density based on lower-index convention

Tilde

Overline

Underline

“”

”

Page 26

26

[25] J. N. Goldberg. Strong conservation laws and equations of motion in covariant field theories. Phys. Rev., 89:263, 1953.

[26] H. J. Meister and A. Papapetrou. Die Bewegungsgleichungen in der allgemeinen Relativit¨ atstheorie und die Koordinatenbe-

dingung. Bull. Acad. Pol. Sci., 3:163, 1955.

[27] L. Infeld. Equations of motion in general relativity theory and the action principle. Rev. Mod. Phys., 29:398, 1957.

[28] R. P. Kerr. The Lorentz-covariant approximation method in General Relativity I. Nuovo Cimento, 13:469, 1959.

[29] R. P. Kerr. The Lorentz-covariant approximation method in General Relativity II - Second approximation. Nuovo Cimento,

13:492, 1959.

[30] V. A. Fock. The theory of space time and gravitation. Pergamon Press, New York (Orig. 1955), 1959.

[31] W. Tulczyjew. Motion of multipole particles in general relativity theory. Acta Phys. Pol., 18:393, 1959.

[32] L. Infeld and J. Plebanski. Motion and Relativity. Pergamon Press, New York, 1960.

[33] R. P. Kerr. On the quasi-static approximation in General Relativity. Nuovo Cimento, 16:26, 1960.

[34] J. L. Synge. Relativity: The general theory. North-Holland, Amsterdam, 1960.

[35] P. Havas and J. N. Goldberg. Lorentz-invariant equations of motion of point masses in the general theory of relativity.

Phys. Rev., 128:398, 1962.

[36] B. Tulczyjew and W. Tulczyjew. On multipole formalism in General Relativity. Recent Developments in General Relativity,

Polish Scientific Publishers, Warsaw, page 465, 1962.

[37] A. H. Taub. Motion of test bodies in General Relativity. J. Math. Phys. (N.Y.), 5:112, 1964.

[38] W. G. Dixon. A covariant multipole formalism for extended test bodies in General Relativity. Nuovo Cimento, 34:317,

1964.

[39] P. Havas. The connection between conservation laws and laws of motion in affine spaces. J. Math. Phys. (N.Y.), 5:373,

1964.

[40] J. Madore. The equations of motion of an extended body in General Relativity. Ann. Inst. Henri Poincar´ e., A11:221,

1969.

[41] W. G. Dixon. Dynamics of extended bodies in General Relativity. I. Momentum and angular momentum. Proc. R. Soc.

A, 314:499, 1970.

[42] W. G. Dixon. Dynamics of extended bodies in General Relativity. II. Moments of the charge-current vector. Proc. R. Soc.

A, 319:509, 1970.

[43] D. E. Liebscher. Generalized equations of motion I: The equivalence principle and non-Riemannian space-times. Ann.

Phys. (Leipzig), 485:309, 1973.

[44] D. E. Liebscher. Generalized equations of motion II: The integration of generalized systems of dynamical equations. Ann.

Phys. (Leipzig), 485:321, 1973.

[45] D. E. Liebscher. The motion of test-particles in non-Riemannian space-time. Cosmology and Gravitation: Spin, torsion,

rotation, and supergravity, edited by P.G. Bergmann and V. De Sabbata, NATO Advanced Study Institutes, Plenum Press,

New York, 58:125, 1979.

[46] A. Papapetrou. Lectures on General Relativity. Reidel, Dordrecht, 1974.

[47] W. G. Dixon. Extended bodies in General Relativity: Their description and motion. Isolated gravitating systems in General

Relativity, Proceedings of the International School of Physics “Enrico Fermi,” Course LXVII, edited by J. Ehlers, North

Holland, Amsterdam, page 156, 1979.

[48] P. B. Yasskin and W. R. Stoeger. Propagation equations for test bodies with spin and rotation in theories of gravity with

torsion. Phys. Rev. D, 21:2081, 1980.

[49] I. Bailey and W. Israel. Relativistic dynamics of extended bodies and polarized media: An eccentric approach. Ann. Phys.

(N.Y.), 130:188, 1980.

[50] F. W. Hehl, J. D. McCrea, E. W. Mielke, and Y. Ne’eman. Metric-affine gauge theory of gravity: Field equations, Noether

identities, world spinors, and breaking of dilation invariance. Phys. Rep., 258:1, 1995.

[51] E. Cosserat and F. Cosserat. Th´ eorie des corps d´ eformables. Hermann, Paris, 1909.

[52] J. Weyssenhoff and A. Raabe. Relativistic dynamics of spin-fluids and spin-particles. Acta Phys. Pol., 9:7, 1947.

[53] E. Kr¨ oner. Kontinuumstheorie der Versetzungen und Eigenspannungen. Ergebnisse der angewandten Mathematik, edited

by L. Collatz and F. L¨ osch, Springer-Verlag, Berlin, 5:1 – 179, 1958.

[54] C. Truesdell and R. A. Toupin. The classical field theories. Handbuch der Physik, edited by S. Fl¨ ugge, Springer-Verlag,

Berlin, III/1:226, 1960.

[55] R. D. Mindlin. Micro-structure in linear elasticity. Arch. Ration. Mech. Anal., 16:51, 1964.

[56] G. Capriz. Continua with microstructure. Springer Tracts in Natural Philosophy, Springer-Verlag, Berlin, 1989.

[57] Y. N. Obukhov and V. A. Korotky. The Weyssenhoff fluid in Einstein-Cartan theory. Class. Quantum Grav., 4:1633, 1987.

[58] W. Kopczy´ nski. Lagrangian dynamics of particles and fluids with intrinsic spin in Einstein-Cartan space-time. Phys. Rev.

D, 34:352, 1986.

[59] Y. N. Obukhov and O. B. Piskareva. Spinning fluid in general relativity. Class. Quantum Grav., 6:L15, 1989.

[60] W. Kopczy´ nski. Variational principles for gravity and fluids. Ann. Phys. (N.Y.), 203:308, 1990.

[61] Y. N. Obukhov and R. Tresguerres. Hyperfluid - a model of classical matter with hypermomentum. Phys. Lett. A, 184:17,

1993.

[62] Y. N. Obukhov, V. N. Ponomariev, and V. V. Zhytnikov. Quadratic Poincar´ e gauge theory of gravity: A comparison with

the General Relativity Theory. Gen. Relativ. Gravit., 21:1107, 1989.

[63] Y. N. Obukhov. Poincar´ e gauge theory: Selected topics. Int. J. Geom. Meth. Mod. Phys., 3:95, 2006.

[64] F. Gronwald and F. W. Hehl. On the gauge aspects of gravity. Proceedings of the International School of Cosmology and

Gravitation: 14th Course, Erice, Italy, edited by P.G. Bergmann et al. (World Scientific, Singapore), page 148, 1996.

Page 27

27

[65] W. Adamowicz. Plane waves in gauge theories of gravitation. Gen. Relativ. Gravit., 12:677, 1980.

[66] F. M¨ uller-Hoissen and J. Nitsch. Teleparallelism - a viable theory of gravity? Phys. Rev. D, 28:718, 1983.

[67] M. Q. Chen, D. C. Chern, R. R. Hsu, and W. B. Yeung. Plane-fronted torsion waves in a gravitational gauge theory with

a quadratic Lagrangian. Phys. Rev. D, 28:2094, 1983.

[68] R. Sippel and H. Goenner. Symmetry classes of pp-waves. Gen. Relativ. Gravit., 18:1229, 1986.

[69] P. Singh and J. B. Griffiths. A new class of exact solutions of the vacuum quadratic Poincar´ e gauge field theory. Gen.

Relativ. Gravit., 22:947, 1990.

[70] A. V. Zhytnikov. Wavelike exact solutions of R + R2+ Q2gravity. J. Math. Phys. (N.Y.), 35:6001, 1994.

[71] O. V. Babourova, B. N. Frolov, and E. A. Klimova. Plane torsion waves in quadratic gravitational theories in Riemann-

Cartan space. Class. Quantum Grav., 16:1149, 1999.

[72] R. W. Tucker and C. Wang. Black holes with Weyl charge and non-Riemannian waves. Class. Quantum Grav., 12:2587,

1995.

[73] A. Garc´ ıa, C. L¨ ammerzahl, A. Mac´ ıas, E. W. Mielke, and J. Socorro. Colliding waves in metric-affine gravity. Phys. Rev.

D, 57:3457, 1998.

[74] A. Garc´ ıa, A. Mac´ ıas, D. Puetzfeld, and J. Socorro. Plane-fronted waves in metric-affine gravity. Phys. Rev. D, 62:044021,

2000.

[75] A. Mac´ ıas, C. L¨ ammerzahl, and A. Garc´ ıa. A class of colliding waves in metric-affine gravity, nonmetricity and torsion

shock waves. J. Math. Phys. (N.Y.), 41:6369, 2000.

[76] D. Puetzfeld. An exact-plane fronted wave solution in metric-affine gravity. Exact solutions and scalar fields in gravity:

Recent developments, edited by A. Mac´ ıas, J. Cervantes-Cota, and C. L¨ ammerzahl (Kluwer, Dordrecht), page 141, 2001.

[77] A. D. King and D. Vassiliev. Torsion waves in metric-affine field theory. Class. Quantum Grav., 18:2317, 2001.

[78] V. Pasic and D. Vassiliev. PP-waves with torsion and metric-affine gravity. Class. Quantum Grav., 22:3961, 2005.

[79] Y. N. Obukhov. Plane waves in metric-affine gravity. Phys. Rev. D, 73:024025, 2006.

[80] P. Baekler, N. Boulanger, and F. W. Hehl. Linear connections with a propagating spin-3 field in gravity. Phys. Rev. D,

74:125009, 2006.

[81] F. W. Hehl and A. Mac´ ıas. Metric-affine gauge theory of gravity II. Exact solutions. Int. J. Mod. Phys. D, 8:399, 1999.

[82] D. Puetzfeld. Status of non-Riemannian cosmology. New Astron. Rev., 49:59, 2005.

[83] P. Baekler and F. W. Hehl. Rotating black holes in metric-affine gravity. Int. J. Mod. Phys. D, 15:635, 2006.

[84] Y. N. Obukhov and G. F. Rubilar. Invariant conserved currents in gravity theories with local Lorentz and diffeomorphism

symmetry. Phys. Rev. D, 74:064002, 2006.

[85] F. P. Chen. Momentum, angular momentum, and equations of motion for test bodies in space-time with torsion. Int. J.

Theor. Phys., 32:373, 1993.

[86] F. W. Hehl. How does one measure torsion of space-time? Phys. Lett. A, 36:225, 1971.

[87] A. Trautman. On the Einstein-Cartan equations III. Bull. Acad. Pol. Sci., s´ er. sci. math. astr. phys., 20:895, 1972.

[88] W. R. Stoeger and P. B. Yasskin. Can a macroscopic gyroscope feel torsion? Gen. Relativ. Gravit., 11:427, 1979.

[89] Y. Ne’eman and F. W. Hehl. Test matter in a spacetime with nonmetricity. Class. Quantum Grav., 14:A251, 1997.

[90] 2007. URL http://einstein.stanford.edu/.

[91] H. Weyl. Gravitation und Elektrizit¨ at. Sitzungsb. Preuss. Akad. Wiss., page 465, 1918.

[92] H. Weyl. Eine neue Erweiterung der Relativit¨ atstheorie. Ann. Phys. (Leipzig), 364:101, 1919.

[93] F. W. Hehl, P. von der Heyde, G. D. Kerlick, and J. M. Nester. General relativity with spin and torsion: Foundations and

prospects. Rev. Mod. Phys., 48:393, 1976.

[94] W. Beiglb¨ ock. The center-of-mass in Einstein’s theory of gravitation. Commun. Math. Phys., 5:106, 1967.

[95] J. Frenkel. Die Elektrodynamik des rotierenden Elektrons. Z. Phys., 37:243, 1926.

[96] E. Corinaldesi and A. Papapetrou. Spinning test-particles in General Relativity. II. Proc. Roy. Soc. London Ser. A: Math.

Phys. Sci., 209:259, 1951.

[97] F. A. E. Pirani. On the physical significance of the Riemann tensor. Acta Phys. Pol., 15:389, 1956.

[98] A. H. Taub. The motion of multipoles in General Relativity.

on General Relativity, Problems of Energy and Gravitational Waves, edited by G. Barb` era, Comitato Nazionale per le

Manifestazioni Celebrative, Florence, page 100, 1965.

Proceedings of the Galileo Galilei Centenary Meeting