1. Introduction
This paper continues the development of a unified perspective on Poincaré and Galilei relativity begun in the first paper in this series (hereafter Paper I) [
1]. As used here, ‘relativity’ refers to the invariance of physical laws under the action of a symmetry group that mixes time and space. The Poincaré group mixes time into space and space into time, whereas the Galilei group only mixes time into space while leaving time invariant. The profound consequences of the Poincaré group’s mixing of space into time—such as the dependence of the time interval between two events on the motion of the observer—are the phenomena traditionally labeled ‘relativistic’. However, the dependence of the space interval between two events on the motion of the observer also renders physics governed by the Galilei group ‘relativistic’ as far as space is concerned. Therefore, Paper I begins by arguing that the traditional terms ‘non-relativistic physics’ and ‘relativistic physics’ should be replaced by the more precise terms ‘Galilei physics’ (or ‘Galilei relativity’) and ‘Poincaré physics’ (or ‘Poincaré relativity’), respectively.
With physics governed by the Poincaré group or the Galilei group both being ‘relativistic’ in this sense, Paper I argues further that the terms ‘special relativity’ and ‘general relativity’ ideally would be released from their traditional association with ‘physics according to Einstein’ to denote more generally ‘physics on flat spacetime’ and ‘physics on curved spacetime’ respectively, regardless of whether physics is governed by the Galilei group or the Poincaré group—globally in the case of flat spacetime, but only locally in the case of curved spacetime. Thus, on flat spacetime, one would speak of both ‘Galilei special relativity’ and ‘Poincaré special relativity’, with the latter being ‘special relativity’ as traditionally understood following Einstein. On curved spacetime, one similarly might consider a ‘Galilei general relativity’ alongside ‘Poincaré general relativity’, with the latter being ‘general relativity’ as traditionally understood following Einstein.
The relativistic invariance of Poincaré physics is manifest, indeed automatic, when expressed in terms of equations governing tensor fields on 4-dimensional spacetime. Einstein’s proposal of Poincaré special relativity as the solution to theoretical and empirical puzzles posed by Maxwell’s electrodynamics was formulated in terms of time-dependent fields on 3-dimensional position space. Minkowski subsequently introduced the concept of spacetime as a flat 4-dimensional Lorentz manifold (pseudo-Riemann manifold with metric tensor of signature (−, +, +, +)) unifying time and space; this allowed him to express electrodynamics in terms of the electromagnetic 4-potential (a covector or linear form) and electromagnetic field tensor (an antisymmetric bilinear form) unifying the electric and magnetic fields, and material particle dynamics in terms of the 4-momentum, whose vector version unifies mass with vector 3-momentum, and whose covector version unifies energy with covector 3-momentum. In short order, von Laue unified the energy density, energy flux, momentum density, and momentum flux of a material continuum in the energy-momentum 4-flux tensor, and this—along with Minkowski’s introduction of spacetime—was key to Einstein’s development of Poincaré general relativity in order to accommodate gravity as spacetime curvature. (References to the original literature of the early 20th century can be found in historical notes in the relevant sections of [
2].)
One can also try to shoehorn Galilei physics into a 4-dimensional spacetime perspective (e.g., [
3,
4,
5], Paper I, and the historical references therein), but the fit is uneasy and incomplete. The fit is uneasy because the spacetime that arises from the infinite speed of light (
) limit of the Einstein metric and its inverse is not a Lorentz manifold: there is no spacetime metric and, therefore, no metric duality of spacetime tensors (raising and lowering of indices); no Levi-Civita connection on spacetime (natural covariant derivative determined by the metric) and no Levi-Civita tensor on spacetime (natural volume form associated with the metric). The fit is incomplete because of the strict separation of mass and energy implied by Galilei physics: a tensor formalism on 4-dimensional spacetime makes manifest the relativistic invariance of conservation of matter and balance of 3-momentum, but not the balance of energy. A 4-velocity vector and a matter-momentum 4-flux tensor are natural enough, but there is no satisfactory 4-momentum covector or energy-momentum 4-flux tensor. The root of the problem is that transformations of time and space yield corresponding transformations of inertia and 3-vector momentum; only through metric duality do these directly correspond also to transformations of energy and 3-covector momentum. Both Galilei physics and Poincaré physics do include a position space 3-metric that relates 3-vector momentum to 3-covector momentum, so that the balance of 3-momentum can be expressed indifferently in terms of either. And the spacetime metric of Poincaré physics implies the equivalence of mass and energy through which the relativistic invariance of balance of inertia is also the relativistic invariance of balance of energy. But Galilei transformations leave the mass of a material particle invariant, and they do not directly exhibit the transformation of kinetic energy implied by the transformation of its 3-momentum. And for a material continuum, Galilei transformations do not allow the first law of thermodynamics to be integrated into a mass-momentum 4-flux tensor, as happens in Poincaré physics. Without a spacetime metric and in failing to manifestly include energy, the practical and aesthetic appeal of a spacetime tensor formalism is significantly compromised in the case of Galilei physics on a 4-dimensional spacetime manifold.
However, a tensor formalism for Galilei physics that manifestly exhibits the transformation of kinetic energy can be accommodated in a 5-dimensional spacetime setting governed by the Bargmann group, a central extension of the Galilei group (e.g., [
3,
6,
7], Paper I, and the references therein). The Bargmann group emerged in connection with representation theory in quantum mechanics, with the Galilei group as a notable example ([
8], see also, e.g., [
9,
10,
11]). Still with a focus on quantum physics, the Bargmann group inspired a tensor formalism for Galilei physics on a 5-dimensional extended spacetime [
12,
13], and its relevance for classical (i.e., non-quantum) Galilei physics was subsequently recognized as well [
14]. It is worth emphasizing at the outset that in this well-established reformulation of Galilei relativity, the extended spacetime introduces no new physical degrees of freedom and serves only as a mathematical device to better express Galilei physics in a tensor formalism.
In part—but only in part—the move from a 4-dimensional spacetime setting to a 5-dimensional extended spacetime setting is conceptually similar to the move from a 3-dimensional position space setting to a 4-dimensional spacetime setting. Consider a classical material particle with no internal degrees of freedom, described completely by its position in 3-dimensional position space as a function of time t according to a fiducial (that is, Eulerian or ‘lab frame’) observer. These points trace out the particle’s trajectory in position space, a (not necessarily injective, or 1-to-1) curve parametrized by t with tangent vector field , the coordinate 3-velocity. Described in terms of 4-dimensional spacetime, the particle’s location according to a fiducial observer can be given as a function of the proper time measured by a comoving (that is, Lagrangian or ‘material frame’) observer moving along with the particle. These points in spacetime trace the particle’s worldline, a (now definitely injective) curve parametrized by with tangent vector field , the 4-velocity. But while the particle is now regarded as a ‘history’ in 4-dimensional spacetime, from a kinematical and dynamical perspective it is still characterized by only three degrees of freedom. Thus, the 4-velocity , and the worldline it determines, are subject to a constraint: in the case of Poincaré physics, where is the Einstein metric; or in the case of Galilei physics, where (not to be confused with proper time , a scalar) is the time 1-form normal to the spacelike position-space leaves associated with absolute time. In a 5-dimensional extended spacetime motivated by the Bargmann group, the particle location now includes an additional ‘action coordinate’ related to kinetic energy per unit mass, in such a way that, not only for Poincaré physics but now also for Galilei physics, the extended spacetime is a pseudo-Riemann (indeed, Lorentz) manifold with metric . In this new setting, the forms (for Poincaré physics) and (for Galilei physics) are still invariant structures governing causality, and the same constraint or applies, where is the 5-velocity tangent to the particle worldline in Bargmann-extended spacetime. The Bargmann metric, , governs the extended spacetime geometry, and it also provides an additional constraint, , ensuring that the particle continues to be characterized by only three degrees of freedom. But unlike the time coordinate t associated with the move to 4-dimensional spacetime, the additional coordinate associated with the move to extended 5-dimensional spacetime is afforded no independent physical significance, and no explicit field dependence on it is allowed (all partial derivatives with respect to vanish). The utility of the additional dimension in the present context is purely to allow Galilei physics to be expressed in a manifestly invariant formalism in terms of spacetime tensor fields on a pseudo-Riemann manifold.
The purpose of this series is to develop a more unified perspective on Poincaré and Galilei relativity, including by exploring the possibility of a strong-field Galilei general relativity that could serve as a useful approximation in astrophysical scenarios such as core-collapse supernovae. The basic strategy is to reexpress standard Poincaré physics on 4-dimensional spacetime in a 5-dimensional setting more congenial to Galilei physics and then deduce the corresponding Galilei-invariant theory by taking the limit. Focusing on flat spacetime, Paper I took a first step by elucidating and emphasizing the fact that the Poincaré group can also be centrally extended to what might be called the Bargmann-Poincaré group, in a manner analogous to the more familiar central extension of the Galilei group to what might now be called the Bargmann–Galilei group (traditionally simply the ‘Bargmann group’).
This installment in the series picks up where Paper I left off:
Section 2 (re)introduces a curved spacetime version of the Bargmann metric
, derived using a procedure similar to that employed in Paper I in flat spacetime, but using the 1 + 3 (traditionally ‘3 + 1’) formalism of Poincaré general relativity as the starting point. (Useful as it may be for the purpose of numerically solving the Einstein equations as an initial value problem, a 1 + 3 spacetime foliation, of course, is not fundamental to a spacetime perspective on Poincaré physics. But the absolute time of Galilei physics does require a 1 + 3 spacetime foliation, making it necessary as a common setting that enables a more unified perspective on Poincaré and Galilei physics.) The projection operator
needed for 1 + 3 and 1 + 3 + 1 tensor decompositions is the subject of
Section 3, after which
Section 4 relates the spacetime Levi-Civita connections
∇ (associated with the 4-metric
) and
(associated with the 5-metric
) to the position space Levi-Civita connection
associated with the 3-metric
.
The gravitational ‘kinematics’ referred to in the title of this paper comes into focus in
Section 5, where it is demonstrated that the extrinsic curvature tensor
of the usual 1 + 3 formalism of Poincaré relativity (e.g., [
15]) is all that is needed to also completely characterize the extrinsic geometry of the position space spacelike leaves in the 1 + 3 + 1 foliation of the Bargmann spacetimes considered here. The label ‘kinematics’ refers to the fact that the usual 1 + 3 formalism of Poincaré relativity locates the gravitational degrees of freedom in the 3-metric
, with the extrinsic curvature
describing the evolution of
between neighboring position space leaves, thus constituting a kind of ‘velocity’ of the gravitational degrees of freedom. The gravitational ‘dynamics’—the Einstein equations relating the spacetime metric to the energy-momentum content on spacetime, and the evolution of
emerging therefrom (in effect, the ‘acceleration’ of the gravitational degrees of freedom)—will be addressed in the next paper in this series.
The remaining sections in the present installment serve as preparation for the gravitational dynamics to be considered in the sequel, and they also provide an initial application of the geometry of the curved Bargmann spacetimes elucidated here to the motion of an elementary material particle and a simple fluid. As mentioned above, the initial move in the overarching strategy employed in this series is to re-express standard Poincaré physics on 4-dimensional spacetime in a 5-dimensional setting more congenial to Galilei physics. Key to this ‘encoding’ of familiar 4-dimensional physics in a 5-dimensional setting is a ‘decoding’ operator
introduced in
Section 6. Its utility for obtaining tensor laws on 5-dimensional Bargmann spacetime from tensor laws on 4-dimensional spacetime is illustrated in
Section 7 on the dynamics of an elementary particle and
Section 8 on the dynamics of a simple fluid. The latter section introduces the kinetic-energy–momentum–mass-density 5-flux tensor
, the 5-dimensional encoding of the total-energy–momentum 4-flux tensor
; this can be expected to appear in the Einstein equations on the 5-dimensional Bargmann spacetimes to be considered in a subsequent installment. Along with a concluding summary,
Section 9 includes remarks about the relationships between this work and previous instantiations of Newton gravity in a 5-dimensional setting [
3,
7,
14], and generalizations of Newton-Cartan gravity in 4-dimensional spacetime in which the connection includes torsion (e.g., [
16,
17,
18,
19,
20,
21]).
6. Obtaining Physical Laws on Bargmann Spacetimes
The presentation thus far has considered the 4-metric on an Einstein spacetime , and the 5-metric on a Bargmann spacetime or ; the decompositions of these and other tensors according to a foliation of spacetime into spacelike position space leaves with 3-metric ; and multiple interpretations of the extrinsic curvature tensor , which has been shown here to be the same on or as on . Apart from a few allusions to the motion of material particles, for the most part it is geometry that has been discussed to this point, in particular aspects related to spacetime foliations. Physics does not really enter the picture until physical laws are given in the form of tensor equations on spacetime.
How are physical laws embodying a unified perspective on Poincaré and Galilei physics on curved Bargmann spacetimes
and
to be obtained? Recall the general strategy mentioned in
Section 1: first, re-express a known physical law on the usual Einstein spacetime
as a law with the same physical content but in a form appropriate to Bargmann–Einstein spacetime
; and second, consider a
limit of that law appropriate to the closely related Bargmann–Galilei spacetime
. The first step of this procedure—the
encoding on
of known Poincaré physics on
—might be accomplished by reverse-engineering equations that express the
decoding of physics on
back to physics on
. Such decoding expressions involve an operator
that relates tensors on
to tensors on
, in a manner characterized by important comparisons and contrasts with the operators
in Equations (
27) and (
34), which respectively project tensors from spacetime
or
to a position space slice
.
In order to understand these comparisons and contrasts, it is helpful first to review the several relationships between tensors on spacetimes or and tensors on a leaf of the spacetime foliation; this review then fosters a working understanding of the relationships between tensors on spacetime and tensors on spacetime . The entire matter is unlocked through the following key insight: the structure of a Bargmann spacetime is such that it furnishes relationships between tensors on and that are partly analogous to those associated with the embedding of the leaves in or , but with the roles of tensors (including vectors) and tensors (including 1-forms) reversed.
Following the elucidation of these relationships and some properties of the decoding operator , a proposed procedure for reverse engineering physical laws on Einstein spacetime to yield tensor relations on Bargmann–Einstein spacetime is described and illustrated with an elementary example.
6.1. Relationships Between Tensors on Spacetime or and Tensors on Position Space Leaves
Here the several relationships between tensors on spacetimes
or
and tensors on a leaf
of the spacetime foliation will be recalled. Begin with relationships that exist naturally by virtue of the embedding of
in
or
: the
push-forward of a vector field (or, more generally, any
tensor field) from
to
or
, and the
pull-back of a 1-form (or, more generally, any
tensor field) from
or
to
. Then, consider the relationships in the opposite directions, which differ in that they make explicit reference to 1-forms and vector fields normal to
. These relations are the
projection of a vector field (or, more generally, any
tensor field) from
or
to
, and the
extension of a 1-form (or, more generally, any
tensor field) from
to
or
. They exist thanks to the operator
in Equations (
25) and (
32). It will prove consistent to adopt a naming convention in which tensors that originate on
are denoted with the
same symbol when they are regarded as or extended to tensors on
or
, while tensors induced on or projected to
that originated on
or
are given a
different symbol.
Consider first the push-forward of a vector field
on
to the spacetime
or
. Note that
may be regarded either in isolation as a manifold in its own right or as a submanifold embedded in
or
. As a vector field on
regarded in isolation as a manifold in its own right, at any point,
gives the tangent vector to some curve in
, and it is represented by a 3-column
. When
is regarded as a submanifold embedded in
or
, that curve in
and its tangent vector are now a curve that still lies in the submanifold
and a vector in
or
that now happens to be tangent to the submanifold
. In coordinates adapted to the foliation,
regarded as a vector in spacetime is now represented by a 4-column (for
) or 5-column (for
) with additional vanishing components, since the vector does not ‘point away’ from
:
The convention is that the same symbol (in this case,
) is used to denote the vector field on
in isolation and its push-forward to a vector field on
or
that happens to be tangent to
embedded in
or
.
Consider next the pull-back of a 1-form
from
or
to a leaf
of the foliation. At any point,
is a real-valued function of vectors tangent to
or
, and is represented by a 4-row
(on
) or 5-row
(on
). The pull-back of
to
amounts to the restriction of its domain to vectors tangent to
. As
may be considered in isolation as a manifold in its own right, it is conventional to denote this restricted function as a 1-form on
using a different symbol—say, for instance,
in the present example—even though its components are precisely the same as the position space components of
with respect to coordinates adapted to the foliation:
That is, in coordinates adapted to the foliation, while the push-forward of a vector involves simply padding the time and action components with zeros, the pull-back of a form involves simply deleting the time and action components. The obvious example of a pull-back of a form in the present work is the 3-metric
on
, whose components with respect to coordinates adapted to the foliation are simply
(for
) or
(for
). This is what has been meant by prior statements that the Einstein metric
or the Bargmann metric
on spacetime
induces a 3-metric
on the leaves of the foliation.
Turn now to the first ‘opposite direction’ operation, the projection of a vector field from
or
to a leaf
of the foliation. A vector field
on
or
, can be decomposed into pieces tangent to
and normal to
:
The vector field
is represented as the 4-column or 5-column
That is, use of the projection operator
given by Equations (
25) or (
32) yields the vector field
tangent to
. When
is regarded in isolation as a manifold in its own right,
is represented by the 3-column
. But when
is regarded as a submanifold of spacetime,
is represented by the 4-column or 5-column
In general, the projection to
involves ‘information loss’; and as illustrated here, the convention is that different symbols are used to denote the original vector field in spacetime and its projection to a leaf of the foliation, unless the original vector field in spacetime already happens to be tangent to
.
Turn finally to the second ‘opposite direction’ operation, the extension of a 1-form on
to
or
. A 1-form
on
regarded in isolation as a manifold in its own right is represented by a 3-row
. Regarded as a function on vectors, an extension of
to an expanded domain beyond vectors on
requires additional information (in contrast to the case of projection, which generally deletes information). In the present case of extending
to
or
when
is regarded as a submanifold, the ‘extra information’ is simply that the value of
vanishes for vectors normal to
. This is accomplished by writing
with
on the left the extension to
or
and the expression on the right the original tensor on
, with
guaranteeing only vector arguments tangent to
. Another way to think about this is to imagine
on
being raised to the 3-vector field
, pushed forward to
or
(acquiring vanishing time and action components), and then lowered back to a 1-form with the spacetime metric, either
or
. On
for example, in components this sequence of operations corresponds to
. Thus (see the components of
in Equations (
25) and (
32)), the extension is represented by the 4-row or 5-row
The acquired non-vanishing time component ensures that
, consistent with the extended
remaining tangent to
. Note that the convention is that the same symbol is used for both the original 1-form on
and its extension to
or
.
6.2. Relationships Between Tensors on Spacetime and Tensors on Spacetime
With this review of the relationships between tensors on spacetimes or and tensors on a leaf of the spacetime foliation in mind, a comparison and a contrast with the relationships between tensors on spacetime and tensors on spacetime are in order. As mentioned above, the structure of a Bargmann spacetime is such that it furnishes relationships between tensors on and that are partly analogous to those described above associated with the embedding of the leaves in or , but with the roles of tensors (including vectors) and tensors (including 1-forms) reversed.
Consider first a push-forward type of relationship. As described above, vector fields (and tensor fields generally) on can also be regarded as tensor fields on or , and in practical terms, they are placed into this new setting by adding vanishing time components, and also action components in the case of . But as emphasized in Paper I, the nature of the Bargmann groups is such that 1-forms (and tensor fields generally, that is, multilinear forms) on 4-dimensional spacetime are treated in precisely this way, that is, set directly in 5-dimensional Bargmann spacetime with vanishing action components, such that they retain their identity as the ‘same’ tensors in the new setting. Important examples from Paper I include the Einstein metric on (and here, ) and the time form on , which retain their causality-governing function in the Bargmann spacetimes and (and here, the curved Bargmann spacetimes and ). The 2-forms of electrodynamics are additional examples from Paper I. Notice the retention of the same symbol to denote the ‘same tensor’.
Consider next a pull-back type of relationship. As described above, 1-forms (and tensor fields generally) on or induce corresponding covariant tensor fields on by restricting their domain to vectors tangent to , and in practical terms, the components of these induced tensors are simply the position space components of the originating tensors on or . But the nature of the Bargmann groups is such that vector fields (and tensor fields generally) on 5-dimensional Bargmann spacetime can be treated in precisely this way, that is, as inducing corresponding contravariant tensor fields on 4-dimensional spacetime whose components are simply the traditional spacetime components of the originating tensors on 5-dimensional spacetime. Notable examples include the 5-velocity inducing the 4-velocity , and the inverse 5-metric inducing the inverse 4-metric (Poincaré physics) or degenerate inverse ‘4-metric’ (Galilei physics). Notice the use of a different symbol to denote the induced tensor.
Turn now to a projective type of relationship, which in the partial analogue relating tensor fields on
to tensor fields on
will be called ‘decoding’ rather than ‘projection’. As described above, in the reverse direction of the push-forward operation, vector fields (and
tensor fields generally) on
or
can be projected to
via contraction with the projection tensor
, which is the mixed-index version of the extension of the induced 3-metric
on
back to
or
. It turns out that there exists a ‘decoding operator’
partly analogous to the projection operator
. To begin to see this, notice first that the 5-metric
on
‘encodes’ the push-forward of the 4-metric
on
as follows:
or
a relation confirmed directly from the component expressions in Equation (
1), padded with vanishing action components, and Equations (
16) and (
29). Raising the index yields
where in this context
is the identity tensor on
. If
is said to be the ‘Bargmann encoding’ of
, then
provides the decoding:
as confirmed via direct computation. This is projection in part, in the sense that the action components vanish on the left-hand side (
is a non-invertible operator); but instead of being simply deleted, the action components are recombined with the time components in a particular way, hence the designation ‘decoding operator’ rather than ‘projection operator’. The other fundamental example is the spacetime momentum. Recall that the kinetic-energy–momentum–mass covector
on
is the Bargmann encoding of the push-forward of the total-energy–momentum
on
, separating mass from kinetic energy:
or
as confirmed from Equation (
12), padded with vanishing action component, and Equations (
24) and (
29). Once again,
provides the decoding:
as confirmed via direct computation. Note again the use of a different symbol to denote the projected or decoded tensor.
Turn finally to an extension type of relationship. As described above, in the reverse direction of the pull-back operation, 1-forms (and
tensor fields generally) on
can be extended to
or
by applying the projection operator
to vector arguments before an evaluation of the form, and defining the extension of the 1-form by absorbing the projection operator into the form itself. But the nature of Bargmann groups is such that it is vector fields (and
tensor fields generally) that can be extended from
to
in a similar way using the decoding operator
. This can be conceptualized in a manner analogous to one of the ways extensions are discussed above. Begin with a vector field,
on
, represented by a 4-column
. Its extension to
is obtained by lowering
on
to the 1-form
, pushing it forward to
(acquiring vanishing action components), and then raising it back to the vector field
on
. In components, this sequence of operations corresponds to
. Using the components of
in Equation (
68), along with Equation (
29) for
, the result is that the extension is represented by the 5-column
This is reminiscent of the extension of a 1-form from
to
or
in Equation (
66), in that, unlike the push forward, the extension involves the acquisition of an additional nonzero component. Note again that the same symbol is used to denote the original tensor on
and its extension to
. As an example, consider the extensions of the fiducial observer 4-velocity
and general 4-velocity
from
to
, represented by the 5-column
These differ in the last component from the 5-columns
representing the fiducial observer 5-velocity
and general 5-velocity
, which is why it was necessary to use different symbols as insisted in
Section 2.2: the extension of a vector on
to
is not the same as its Bargmann encoding on
. In fact, the proper relationship between 4-velocities on
and 5-velocities on
is one of decoding, rather than extension:
as can be shown via direct calculation with the above component expressions. (Of course, fiducial observer 4- or 5-velocities correspond to general 4- or 5-velocities for 3-velocity
.) Using Equation (
68) in Equation (
72) yields
and then Equation (
28) gives
thus confirming that a vector version of Equation (
5) on
holds for the extension of the fiducial observer 4-velocity
on
to
.
6.3. Properties of the Decoding Operator
Some relations involving the decoding operator
are worth noting. It is idempotent:
It nullfies
, making
non-invertible:
It preserves
:
It transforms
in a manner reminiscent of Equation (
5):
These relations follow from Equation (
68) for
and the mutual contractions in Equations (
30) and (
31). Moreover, its divergence vanishes,
thanks to the vanishing divergence of
in Equation (
63) and the fact that
in Equation (
62) has no action component.
Similar to Equations (
27) and (
34) for the generalized projection operator
for arbitrary tensors, these observations about decoding and extension using
give rise to the generalized decoding operator
that gives the decoding to
of a
tensor
on
.
This generalized decoding operator appears in an important relation between tensor gradients on
and
, similar in spirit to Equation (
45) relating tensor gradients on spacetime to tensor gradients on position space leaves. Consider a tensor
on
that is decoded to
, that is, such that
, so that all possible contractions with
and
vanish. Then, the relation
holds, where
is given by Equation (
77). The reasoning is similar to that leading to Equation (
45): it must be shown that
, and then the uniqueness of the Levi-Civita connection can be invoked to deduce that
. Indeed, it is the case that
thanks to
and Equation (
74).
6.4. Reverse Engineering Poincaré Physics on from Poincaré Physics on
Given physical laws on Einstein spacetime
, the resources and a proposed procedure to be used for encoding them on Bargmann–Einstein spacetime
can now be summarized. First, there are three key encodings that arise in connection with the construction of Bargmann spacetime itself presented in
Section 2.2: the 5-velocity
as the Bargmann encoding of the 4-velocity
, the Bargmann metric
as the Bargmann encoding of the Einstein metric
, and the kinetic-energy–momentum–mass
as the Bargmann encoding of the total-energy–momentum
. (It is worth emphasizing again that the Bargmann encoding
of
is not the same as the extension of
and that the push-forwards of
and
are not the same as their Bargmann encodings
and
.) And second, there is the clear understanding developed above of the relationships between tensors on
and on
: the push-forward of
tensor fields and the extension of
tensors from
to
, and the pull-back of
tensors and decoding of
tensors from
to
. The generalized decoding operator
is key to these relationships. The proposed procedure is to take a physical law on
, use the known Bargmann-encoded entities to write it as the decoding of an expression on
using the decoding operator
, and—with luck or perhaps a bit of additional information—reverse-engineer the resulting decoding to work out the encoding of the physical law on
. What makes the reverse-engineering possible is that the first term of
in Equation (
68) is the identity operator, which allows unprojected tensor expressions on
to appear; the second term of
gives terms that one might hope to explicitly compute.
Consider an elementary example as a warm-up exercise and by way of illustration: What is the Bargmann-encoded version on
of the equation
expressing the normalization of a 4-velocity
on
? It was already noted in
Section 2.2 that the answer is
, but this result is obtained here by using the decoding operator
. The easily-confirmed relation
will be a useful lemma. First, take the equation on
, and set it in
by pushing forward
and extending
. In so doing,
acquires an action component according to Equation (
71), but it contributes nothing because the action components of the push-forward of
vanish:
Next, introduce quantities appropriate to the Bargmann perspective by using Equations (
69) and (
72) to express
and
as the decodings of
and
:
in which the idempotence property of Equation (
73) has been used. Now comes the reverse engineering step, enabled via Equation (
68) for the decoding operator, along with the normalization of
from Equation (
30) and the repeated application of the lemma of Equation (
79):
thus arriving at the expected result,
. Again, notice that it is the identity tensor
in the first term of Equation (
68) for the decoding operator
that enables the desired reverse engineering of physical laws on
from physical laws on
, in this example by allowing the term
to appear.
9. Conclusions
If the consideration of the motion of a material particle in 3-dimensional position space led to Galilei relativity, and investigation of the propagation of light led to Poincaré relativity understood in the context of 4-dimensional spacetime, a retrospective reconsideration of the motion of a material particle leads to a more unified perspective on Poincaré and Galilei relativity on a 5-dimensional extended spacetime—Bargmann-Einstein spacetime
in the case of Poincaré physics, and Bargmann-Galilei spacetime
in the case of Galilei physics. The extra dimension plays no independent physical role implying new degrees of freedom, but, being associated with the kinetic energy per unit mass of a material particle (see Equation (
15)), it enables Galilei physics to be expressed in terms of a spacetime tensor formalism that respects the separation of mass and kinetic energy. This paper builds on Paper I by working this out in curved spacetime, extending the usual
formulation of Poincaré general relativity on Einstein spacetime
to a
setting suitable for both
and
. Indeed, the basic strategy throughout is to translate known Poincaré physics on a curved 4-dimensional spacetime into a 5-dimensional setting, where the corresponding Galilei physics can be deduced by a
limit.
A prime benefit of this ‘Bargmann’ perspective is that the geometry (here, including curvature) of both
and
is governed by a 5-metric
(see Equation (
16)), conferring the several blessings a spacetime metric affords: metric duality of tensors, a Levi-Civita connection, and a Levi-Civita volume form. It is worth reiterating, however, that the forms
(for Poincaré physics) and
(for Galilei physics) that govern causality and the measurement of proper time in a 4-dimensional setting retain this role in the 5-dimensional setting. That the 4-metric
on
governs not only spacetime geometry but causality and time measurement as well, while on
and
these responsibilities are divided between the 5-metric
and either
or
respectively, is one way in which the Bargmann approach requires Poincaré physics to ‘give something up’ for the sake of a more unified perspective yielding greater insight into Galilei physics.
A foundational example of the way the Bargmann perspective enables a tensor formalism for Galilei physics is the unification of energy and momentum (and mass) in a covector or 1-form. As noted in Paper I, a 4-velocity or (multiplying by particle mass) a vector 4-momentum in the form of inertia-momentum, is not a problem for Galilei physics. But taking the metric dual (by
) to obtain the covector or 1-form 4-momentum
—the total-energy–momentum of Equation (
12)—is a feat of Poincaré physics that Galilei physics cannot replicate on a 4-dimensional spacetime. The magic of the Bargmann approach is that taking the metric dual (by
) of the inertia–momentum–kinetic-energy 5-vector
of Equation (
23) yields the kinetic-energy–momentum–mass covector or 1-form
of Equation (
24) in which mass is disentangled from kinetic energy by removing it from the first component and moving it to the fifth component without a factor of
. Of course, that the equivalence up to a factor of
of inertia and total energy is no longer manifest is a second way in which the Bargmann approach requires Poincaré physics to ‘give something up’ for the sake of a more unified perspective, yielding greater insight into Galilei physics.
While physical laws expressed in terms of tensor fields on spacetime embody relativistic invariance, comparison with experiment requires tensor decompositions consistent with the way humans experience time evolution in position space. This is, of course, at the heart of the
formalism on
and the
formalism on
or
featuring a foliation of spacetime into position space leaves. Prominent tensor fields associated with the foliation and tensor decomposition include the 4-velocity field
(on
, see Equation (
7)) or 5-velocity field
(on
or
, see Equation (
21)) of fiducial observers, orthogonal to position space leaves in a timelike direction; and the 1-form
(see Equations (
6) and (
20)) dual to these, in the sense that
and
. An important difference between
on
and
on
or
is manifest in their directional derivatives along themselves (Equations (
37) and (
41), respectively): fiducial observers are accelerated on
, while fiducial observers are geodesic on
or
. On
or
, the vector field
(anti-)parallel to the new action coordinate axis (see Equation (
28)) also points away from the position space leaves, satisfying
but also
. These vector fields and 1-forms appear in the operator
(see Equations (
25) and (
32)) that projects vector fields and 1-forms to the position space leaves. This appears in a generalized projection operator,
, for all tensors on
(see Equation (
27)) or on
or
(see Equation (
34)), including by relating spacetime gradients of tensors tangent to position space leaves to gradients tangent to the leaves (see Equation (
45)).
In this projective relationship between tensor gradients on spacetime and tensor gradients on the position space leaves, the piece—or pieces, in the case of Bargmann spacetimes—that are projected out serve to define extrinsic curvature (see Equations (
46) and (
53)), which is also related to the gravitational ‘kinematics’ referenced in the title of this paper. The label ‘kinematics’ refers to the fact that the standard 1 + 3 formalism of Poincaré general relativity locates the gravitational degrees of freedom in the 3-metric
on the position space leaves, with the extrinsic curvature tensor
serving as a kind of ‘velocity’ of these gravitational degrees of freedom (see Equation (
50)), a relationship confirmed in the 1 + 3 + 1 formalism undertaken here (see Equation (
64)). Meanwhile, the evolution of the projection operator normal to the leaves vanishes (see Equations (
51) and (
65)), with the important consequence that tensors tangent to the leaves remain tangent to the leaves. Turning back to geometry, extrinsic curvature tensors carry information about the way the leaves of a foliation are embedded in the ambient manifold, in that they describe the variation along the leaves of vector fields or 1-forms normal to the leaves; the resulting relations in Equation (
49) and Equations (
61) and (
62) are key to the 1 + 3 and 1 + 3 + 1 decomposition of physical laws expressed in terms of spacetime tensor fields. In the general case of a foliation of a manifold into leaves of codimension 2, one might expect two independent extrinsic curvature tensors corresponding to the two directions normal to the leaves. Remarkably—or perhaps inevitably, in hindsight—even though the position space leaves of
and
are of codimension 2, the geometry of Bargmann spacetimes is constrained in such a way that the two extrinsic curvature tensors are essentially the same (related by a constant factor, see Equation (
59)); and moreover, with
on
or
precisely matching
on
(compare Equations (
55) and (
58) with Equation (
52)). The similarity of Equation (
60) on
or
to Equation (
48) on
, in contrast to the apparent difference between Equations (
53) and (
46), is one manifestation of there effectively being only one extrinsic curvature tensor. This outcome, and the associated consistency of gravitational kinematics on Bargmann spacetimes with that seen in the standard 1 + 3 formulation of Poincaré relativity, is one of the signal results of this paper.
The working out of gravitational ‘dynamics’ in the context of the Bargmann spacetimes and —the constraint equations, and the evolution of the extrinsic curvature , as these result from the Einstein equations on encoded on —is left for the next installment in this series. Therefore, without equations relating the components of on to the energy–momentum–mass content thereon, comparisons of the conjectured strong-field Galilei general relativity with particular applications of Poincaré general relativity are not yet possible; such considerations are outside the scope of this paper, and are left for future work.
But this paper prepares for that next step by proposing a procedure, described in
Section 6, through which known physical laws on
can be translated to
, which might be hoped or expected to be amenable to a
limit yielding physical laws on
. The procedure involves a ‘decoding’ operator
(see Equation (
68)) and its generalization
(see Equation (
77)) relating tensors on
to tensors on
. They are partly analogous to the projection operators
and
discussed previously, which project tensor fields on spacetime to the position space leaves. The basic idea is to express a known physical law on
as the ‘decoding’ of an expression on
and then ‘reverse engineer’ this expression to obtain the ‘encoding’ of this physical law on
. The term ‘decoding’, rather than ‘projection’, is coined because action components are not simply deleted in going from
back to
, but recombined with the time component—a reversal of the kind of separation effected in the Bargmann approach of, for instance, mass from total energy, or the lapse function from the 4-metric.
By way of example, and as an application of the curved Bargmann spacetime geometry proposed here, assuming that the lapse function , shift vector , 3-metric , and extrinsic curvature associated with the 4-metric and 5-metric are given, this reverse engineering procedure is applied to physical laws for two systems: the dynamics of an elementary particle, and the dynamics of a simple fluid.
For the dynamics of a material particle, the physical law in terms of spacetime tensors can be expressed as Equation (
81) on
and by Equation (
88) on
or
. On the right-hand side in the latter case, Equation (
87), inspired by (but by no means limited to) the weak-field limit, has been employed. The
decomposition on
and the
decomposition on
give the same results (in a simpler formulation than that presented, for instance, in [
22]), confirming the physical equivalence of the Bargmann encoding of this dynamical law. It also gives sensible results for
without a restriction on the strength of the gravitational fields, although it appears that retaining nonlinear terms would require a limit in which also
as
in such a way that
remains constant. But despite matching results for the decomposed equations, the spacetime tensor laws of Equations (
81) and (
88) reflect very different perspectives: in referencing
on the right-hand side, the Bargmann-encoded Equation (
88) reverts the Einstein perspective of accelerated fiducial observers (see Equation (
37)) and geodesic material particles (see Equation (
80)) to a Newton-like perspective of geodesic fiducial observers (see Equation (
41))—analogous to Newton’s inertial observers—and accelerated material particles subject to a gravitational force. In the case of Poincaré physics, strictly speaking, Equation (
88) should not be regarded as an invariant spacetime tensor law, since
arises from the foliation which is supposed to be freely chosen by virtue of coordinate freedom. This would be a third way in which the Bargmann approach requires Poincaré physics to ‘give something up’ for the sake of a more unified perspective yielding greater insight into Galilei physics. But in the case of Galilei physics with expectations of absolute time, it would not be surprising for the foliation to turn out to be fixed, such that Equation (
88) actually is an invariant spacetime tensor law.
The other example concerns the dynamics of a simple fluid composed of a single type of microscopic particle of constant mass, with the 4-fluxes (on
) or 5-fluxes (on
or
) characterizing the fluid given by momentum moments of a scalar distribution function. The particle number flux
on
(see Equation (
92)) and
on
or
(see Equation (
113)), and the
and
decompositions (see Equations (
98) and (
116)) of their vanishing divergences (see Equations (
97) and (
115)) are essentially kinematical in nature, in that they define and describe the fluid and its motion. They basically are to the fluid what the definition of 4-velocity
(on
) or 5-velocity
, and the trajectory relation (Equations (
82) and (
89)), are to a material particle; indeed, the particle flux defines the fluid 4-velocity or 5-velocity, the proportionality factor being the particle density measured by a comoving observer. The fluid dynamics is given in terms of the 4-momentum flux
on
(see Equation (
99)) and 5-momentum flux
on
or
(see Equation (
117)); the latter might be expected to appear in the encoding to
of the Einstein equations on
, to be considered in the next installment in this series. The spacetime dynamical law—the divergence of these
tensor fields, given by Equation (
105) on
and by Equation (
123) on
or
—is analogous to Equations (
81) and (
88) for a material particle mentioned previously. Indeed, the gravitational force appearing on the right-hand side of Equation (
123) on
or
is like that for a material particle in Equation (
88). And as was the case for a material particle, the
decomposition on
and the
decomposition on
give the same results, again confirming the physical equivalence of the Bargmann encoding of this dynamical law.
A few words are in order on how the approach taken here differs from or relates to previous work. Previous work in a 5-dimensional spacetime setting has allowed weak-field Newton gravity to be incorporated into the Levi-Civita connection associated with a spacetime 5-metric
[
3,
7,
14]. In the schematic diagram in Figure 1 of Paper I, the spacetime of that theory is denoted
; and based on a comparison of their differing 5-metrics, the spacetime
being explored in this series is indicated in that figure to be qualitatively different.
More can now be said about an important distinction between the weak-field, linear gravitation of
and the potentially strong-field, nonlinear gravitation of
. There is freedom in the choice of spacetime connection (covariant derivative) when one generalizes from flat spacetime to curved spacetime. In the mathematical language of the reduction of a frame bundle, a spacetime symmetry group—here, the Poincaré group or Galilei group—acts ‘vertically’ within each fiber of the frame bundle (relating bases of the spacetime tangent space at a single point of spacetime), while the connection acts ‘horizontally’ (relating bases at neighboring spacetime points). A natural choice is to constrain the connection by requiring that it be ‘compatible’ with tensors that are invariant under the action of the symmetry group, in the sense of requiring that their covariant derivatives vanish. In the present case, the invariant tensors under the Bargmann-Poincaré and Bargmann-Galilei groups are the metric
and the action vector
, so that compatibility for both would require
(the condition defining a Levi-Civita connection) and also
. This is the choice that leads to
. The spacetime
explored here also features a Levi-Civita connection (
), but the 5-metric derived from particle kinematics in the 1 + 3 formulation of Poincaré relativity yields
; see Equation (
62). The conjecture here is that this relaxation of a ‘compatibility’ requirement on
may allow a Levi-Civita connection in the 5-dimensional setting to embody strong-field Galilei gravitation.
This may turn out to be related to work generalizing standard weak-field Newton–Cartan Galilei gravitation on 4-dimensional spacetime to strong-field Galilei gravitation by allowing the connection—not a Levi-Civita connection, since there is no spacetime metric—to include torsion (e.g., [
16,
17,
18,
19,
20,
21]). The Frobenius condition for a 4-dimensional spacetime to be foliated into a family of spacelike hypersurfaces is
, where
is the time form mentioned above. Standard weak-field Newton-Cartan theory uses a torsion-free connection, which happens to be equivalent to
[
23]. This is, of course, stronger than necessary to satisfy the Frobenius condition. In the works cited above, the requirement that the connection be torsion-free is relaxed to allow so-called ‘twistless torsion’ compatible with the Frobenius condition. Whether this is directly related to the present work is not explored here, but it is intriguing to note that the antisymmetry of the right-hand side of Equation (
62), together with the fact that
on
, gives Equation (
62) a rotational (and, in that sense, ‘torsional’) character.
As noted briefly in
Section 1, a strong-field Galilei general relativity could serve as a useful approximation in astrophysical scenarios such as core-collapse supernovae. In this system, gravity associated with the nascent neutron star is enhanced at the 10–20% level by energy density and pressure, along with nonlinearity; perhaps this could be accommodated while enjoying the simplifications of setting aside ‘Minkowski’ bulk fluid flow and the back-reaction of gravitational radiation. (This approach might conceptualized as ‘microscopically Poincaré’ but ‘macroscopically Galilei’.) And while it remains to be seen what the equations governing curvature (the ‘Einstein equations’) on
turn out to be in the sequel to this paper, strong-field Galilei gravitation encoded in twistless torsion on 4-dimensional spacetime gives an indication of what might be expected. For example, the
but strong-field formalism of [
17] is argued in [
18] to be a low-speed expansion about the ‘static sector’ of Poincaré general relativity, in effect a resummation with respect to gravitational strength of the usual (weak-field) post-Newtonian series. While not expressed exactly this way in those works, an inspection of the strong-field but ‘Galilei’ reinterpretation of the Schwarzschild geometry in [
17], and the strong-field but ‘Galilei’ expansion of the Kerr geometry and Oppenheimer–Snyder collapse in [
18], shows that indeed they involve limits in which both
and
such that
remains constant, as deduced here.