See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/354054557

Relative binary, ternary and pseudo-binary 4D velocities in the Special

Relativity: Part I. Covariant Lorentz transformation

Conference Paper · August 2021

CITATION READS

1 124

1 author:

Grzegorz M. Koczan
Warsaw University of Life Sciences - SGGW
77 PUBLICATIONS 139 CITATIONS

SEE PROFILE

All content following this page was uploaded by Grzegorz M. Koczan on 02 March 2022.

The user has requested enhancement of the downloaded file.

 Relative binary, ternary and pseudo-binary
4D velocities in the Special Relativity
Part I. Covariant Lorentz transformation
arXiv:2108.10725v3 [physics.gen-ph] 23 Feb 2022

Grzegorz Marcin Koczan

Warsaw University of Life Sciences (PhD),

Faculty of Physics at the University of Warsaw (MSc)
[email protected], [email protected]

Abstract. Zbigniew Oziewicz was a pioneer of the 4D space-time approach to

covariant relative velocities. In the 1980s (according to private correspondence)
he discovered two types of 4D relative velocities: binary and ternary, along with
the rules for adding them. They were first published in conference materials in
2004, and the second time in a peer-reviewed journal in 2007. These physically
logical and mathematically precise concepts are so subtle that Oziewicz’s numer-
ous preprints have yet to receive the recognition they deserve.
Part I of the work was planned to be a more review, but a thorough review of
the little-known results was made in an original synthetic manner with numerous
generalizations. The Part I presents the Oziewicz–Świerk–Bolós (and Matolcsi or
Bini–Carini–Jantzen) binary relative velocity and the Oziewicz–Ungar–Dragan
(also Celakoska–Chakmakov–Petrushevski on the basis of Urbantke, as well as
Wyk) ternary relative velocity. The Einstein–Oziewicz and Einstein–Minkowski
velocities, which are a four-dimensional generalization of Einstein velocities ad-
dition, also have a ternary character (or at least pseudo-binary). It turns out that
the Lorentz transformation of velocity itself is already ternary anchored. This fact
is explicitly revealed by the covariant Lorentz transformation of velocity, which
is the main tool in Part I of the work.
Part II of the work will be fully original and introduces two 4D ternary general-
izations of relative binary velocity called pseudo-binary relative velocities: cross
and axial. A longer description of Part II follows after the conclusion of Part I of
the work.

Keywords: Ternary Special Relativity (TSR), 4D Einstein–Oziewicz relative ve-

locity, 4D Einstein–Minkowski relative velocity, covariant Lorentz transformation

The fundamental property of velocities is the ability to add them.

Zbigniew Oziewicz 2020

Introduction One of the pillars of Special Relativity (SR) is Einstein’s principle of rel-
ativity and the related postulate of the Lorentzian covariance of physics equations. Un-
fortunately, the postulate of covariance is often incorrectly called invariance. Although
the covariance requirement is widely acknowledged and emphasized, fundamental equa-
tions such as the Lorentz transformation and velocities composition are relieved of this
2 Grzegorz M. Koczan, GOL 2021 conference in memoriam Z. Oziewicz, Mexico

requirement. On the one hand, this condition results from the lack of popularization
of the results of the covariant Lorentz transformation obtained in the last century by
Bażański in 1965 [2], Krause in 1977 and 1978 [22,23] and moreover by Fahnline in
1982 [13]. On the other hand, this state of science is due to a misunderstanding and ig-
norance of the dependence of the Lorentz transformation on some additional direction
in space-time. This direction, in the form of a time-like vector, appears explicitly in the
aforementioned works by Krause and Fahnline, but is exposed to erroneous censorship
in terms of an Æther that violates the principle of relativity. In this way, we have come
to the bizarre apparent contradiction that the covariant Lorentz transformation breaks
the basis on which it is based – the principle of relativity.
The explanation of this paradox and the further development of subtleties of a similar
nature is provided by the research program of Professor Zbigniew Oziewicz (see e.g.
[28,31]). This research program will be called Ternary Special Relativity (TSR) here,
as it is based on the realization of the ternary nature of relative velocities in Special
Relativity. It should be emphasized that TSR is a developmental part of the SR and
does not constitute an alternative content to the SR, but deepens its understanding in
a new conceptual language. The concept of TSR was born thanks to the application
of the methodology of the category theory to the kinematics of SR. In simple terms,
it was about morphisms and isomorphisms generated by Lorentz boosts or relativistic
composition of velocities. Although the ternary nature of velocities composition graphs
is a classic example of a category theory approach, this work is written in 4D covariant
algebra. However, both in the covariant approach and in the category theory, the most
overt independence of the choice of the coordinate system is postulated. It should be
noted, however, that TSR in a way separates the choice of coordinates from the choice
of the observer – which doubles the concept of relativity. Therefore, if we have Cartesian
coordinates, the observer and the body or the observer and two bodies, then we can speak
of a ternary system needed to study ternary relations of motion.
The discovery of the ternary nature of the relativistic composition of velocities raised
in Oziewicz a natural question about the possibility of breaking the hegemony of the
Lorentz group down to the level of a particular kind of abstract Æther [30]. I believe
that in the correspondence with Oziewicz it was possible to work out a certain interpre-
tative consensus on this matter. As part of this consensus, I proposed in correspondence
with Oziewicz the term Oziewicz’s Æther for each of the two 2D planes distinguished by
a Lorentz boost. Oziewicz’s spacetime-like Æther is a 2D hyperplane in which Lorentz
boost work, while Oziewicz’s space-like Æther is a 2D surface, which Lorentz boost
does not change. Both such 2D Oziewicz’s Æthers are not any material distinguished
frames of reference, but they constitute a certain subtle distinguishing, which is intro-
duced by the mere Lorentz boost. Therefore, according to the author, TSR does not
break the Lorentz group and the SR itself. Nevertheless, Oziewicz himself expressed
some doubts in this respect [27,29]. These doubts were mainly based on the lack of asso-
ciativity for relativistic velocities addition and the multitude of relative zero velocities.
The lack of associativity of standard relativistic velocities addition is an indisputable
fact [38,16], but it concerns the pseudo-group (groupoid or loop) of velocities, and not
the entire Lorentz group, which includes, next to Lorentz’s boosts, also rotations. Other
aspects of non-associativity in physics were cited in [34].
 Relative binary and ternary 4D velocities in SR: I. Covariant Lorentz transformation 3

This does not mean that the Thomas-Wigner rotation accompanying composition
of non-parallel velocities depreciates the mathematical structure for composition ve-
locities. Oziewicz called this structure a relativistic groupoid [27], which he argued
by the aforementioned multitude of relative zero velocities. Others refer to it as the
Lobachevsky–Einstein velocity space or a gyrogroup. Investigating this structure, how-
ever called, does not constitute a breach of the Lorentz group, whether or not we agree
with the somewhat exotic multi-zero argument or other argument. In other words, ac-
cording to the author, the TSR should be treated as an original and developmental, but
orthodox approach to the SR. Zbigniew Oziewicz, the creator of this program, was look-
ing for new solutions and perhaps he would not literally agree with the above opinion.
Undoubtedly, Oziewicz introduced a new concepts within TSR and SR, and whether or
not we treat TSR as going beyond the SR or not is of secondary importance. As already
suggested, the author of the article classifies TSR as SR without harming any of them.
The only more widely known competitive approach to TSR (including the Oziewicz
relativistic groupoid) is the aforementioned Ungar’s gyrogroup. However, according to
the author, the idea of a gyrogroup is less innovative and less fundamental than TSR.
The gyrogroup is based on the hyperbolic geometry of the relativistic Lobachevsky–
Einstein velocity space (see appendix in [18]), the foundations of which were already
described in 1911 by Varićak, and further developed e.g. by Fock in the 1960s. More-
over, the gyrogroup did not recognize such fundamental concepts as the covariant binary
and ternary relative velocities. Strictly speaking, Ungar in his very extensive research
introduced the equivalent of ternary velocity, but did not recognize its essential role and
ternary character. Ungar’s greatest discovery, as Oziewicz himself emphasized, is the
discovery of the lack of associativity for the relativistic addition of velocities [38]. Both
Ungar and Oziewicz proposed some concepts of a weaker version of associativity as
part of their own approach to the velocities composition problem.
It’s time to define the basics of TSR in 4D covariant vectors algebra. Consider two
material points to which 4D absolute velocities are assigned, respectively: 𝜈 = {𝜈 𝛼 }
and 𝜇 = {𝜇𝛼 }. These are time-like vectors directed to the future, which in the signature
+−−− are normalized in a positive sense to the speed of light 𝜈◦𝜈 = 𝜈 𝛼 𝜈𝛼 = +𝑐 2 = 𝜇◦𝜇.
These will be referred to here as Minkowski 4D absolute velocities or simply Minkowski
velocities. The name four-velocity is commonly used for them, which Oziewicz did not
like. Oziewicz, following Minkowski, thought that the 𝜈 𝛼 , 𝜇𝛼 objects describe material
points (material bodies) as such, and do not constitute their velocities. In a way, he was
right, but in the sense of properly scaled spatial components, Minkowski time-like vec-
tors do describe velocities.
On the basis of Minkowski velocities of two material points, it is possible to build a
general concept of relative velocity – first of the binary type:
𝛼 (𝜈, 𝜇) = 𝜆2 (𝜈, 𝜇)𝜇𝛼 − 𝜆1 (𝜈, 𝜇)𝜈 𝛼 . (1)
Binary velocity, like any relative velocity, is a certain subtraction of component ve-
locities. Such subtraction in relativism requires at least one non-trivial "linear" combi-
nation coefficient. In the case of the canonical Oziewicz–Świerk–Bolós binary veloc-
ity 𝜔𝛼 only the 𝜆2 coefficient is non-trivial. Any binary velocity should be a space-
like vector normalized (in a negative sense) to a speed lower than the speed of light
(𝛼 𝛼 = −𝐵 2 , 0 ≤ 𝐵 < 𝑐).
4 Grzegorz M. Koczan, GOL 2021 conference in memoriam Z. Oziewicz, Mexico

If material points with Minkowski velocities 𝜈 𝛼 , 𝜇𝛼 are observed by an observer with

Minkowski velocity 𝜎 𝛼 , then the relative velocity of the ternary type can be considered:

 𝛼 (𝜎, 𝜈, 𝜇) = 𝑓2 (𝜎, 𝜈, 𝜇)𝜇 𝛼 − 𝑓1 (𝜎, 𝜈, 𝜇)𝜈 𝛼 − 𝑓0 (𝜎, 𝜈, 𝜇)𝜎 𝛼 . (2)

The point is that we want to relate (transfer) 𝜇𝛼 velocity relative to 𝜈 𝛼 to the 𝜎 𝛼 system.
In practice, such a reference requires orthogonality ( 𝛼 𝜎𝛼 = 0), which partially explains
the difference of  𝛼 compared to 𝛼 . The fulfillment of this condition can be guaranteed
automatically by a "linear" combination of canonical binary velocities:

𝑐2 𝛼 𝑐2 𝛼
 𝛼 = 𝑔2 𝜔𝛼2 (𝜎, 𝜇) − 𝑔1 𝜔𝛼1 (𝜎, 𝜈) = 𝑔2 𝜇 − 𝑔1 𝜈 − 𝑔0 𝜎 𝛼 . (3)
𝜎◦𝜇 𝜎◦𝜈
Such a notation only has the coefficients 𝑔0 , 𝑔1 , 𝑔2 scaled with respect to 𝑓0 , 𝑓1 , 𝑓2 , which
simplifies the orthogonality condition to 𝑔0 = 𝑔2 − 𝑔1 . The canonical ternary velocity
of Oziewicz–Ungar–Dragan 𝜉(𝜎) 𝛼 , however, has equal two coefficients of the first kind

𝑓1 = 𝑓2 , not the second kind (𝑔1 and 𝑔2 ). Of course, any ternary velocity should come
down to the binary velocity for 𝜎 𝛼 = 𝜈 𝛼 .
In this work (mainly in Part II) pseudo-binary velocities are also considered. A
pseudo-binary velocity is a ternary velocity in which a principal coefficient is simple,
i.e. unitary (e.g. 𝑔1 = 1 or 𝑔2 = 1), or is a function of only two, not three, Minkowski
velocities. The pseudo-binary velocity will also be a velocity similar to (1), at least one
coefficient of which will depend on the third Minkowski velocity 𝜎 𝛼 . The term pseudo-
binarity refers to ternarity that is not fully manifested, and here it has nothing to do
with the change of signs on reflections (as in pseudo-tensors). The best example of a
pseudo-binary velocity from Part II of the work will be the cross velocity 𝜛 𝛼 . Another
example would be axial velocity 𝜋 𝛼 , which is a 4D generalization of the relative velocity
of Fernándeuz–Guasti (and the author). The pseudo-binarity requirements are also met
by the velocity of Einstein–Oziewicz 𝜀𝛼 generalizing to 4D the Einstein relativistic com-
position of velocity. We assume that pseudo-binary velocity, like binary and ternary, is
a space-like vector (𝜛 𝛼 𝜛𝛼 < 0, 𝜋 𝛼 𝜋𝛼 < 0, 𝜀𝛼 𝜀𝛼 < 0).
An interesting fact is the existence of a time-like (𝛽 𝛼 𝛽𝛼 = +𝑐 2 > 0) relative velocity
of Einstein–Minkowski 𝛽 𝛼 related to 𝜀𝛼 and with the covariant Lorentz transformation.
This velocity is therefore a covariant Lorentz transformation of Minkowski velocity. It
resembles ternary velocity or pseudo-binary velocity, but is a time-like vector.
Some of the following precise conditions are imposed on the relative covariant ve-
locities (binary, ternary, pseudo-binary) occurring in this work:
i. Compliance of velocities composition with the Lorentz transformation (Einstein
composition) for the component parallel to the boost velocity (in the 3D sense).
ii. Compatibility of velocities composition with the Lorentz transformation (Einstein
composition) only in the case of parallel velocities (in the 3D sense).
iii. Corresponding compatibility of the covariant active Lorentz boost of velocity 𝜈 𝛼
performed with some (ternary) relative velocity (agreement in the active sense).
iv. Equality of the 4D relative velocity with the value of some covariant passive Lorentz
boost of velocity 𝜇𝛼 (agreement in the resting sense).
v. Orthogonality to Minkowski velocity reference (𝛼 𝜈𝛼 = 0 or  𝛼 𝜎𝛼 = 0).
 Relative binary and ternary 4D velocities in SR: I. Covariant Lorentz transformation 5

vi. Compatibility with 3D velocity corresponding to 𝜇𝛼 when 𝜈 𝑖 = 0.

vii. The triviality of one of the main coefficients (𝑓1 = 1 or 𝑓2 = 1 or 𝑔1 = 1 or 𝑔2 = 1).
viii. Antisymmetry, i.e. the equality of the main coefficients (𝑓1 = 𝑓2 ).

The work describes the individual relative velocities generally in separate sections.
These sections refer to the above conditions, identifying specific ones and, if necessary,
further specifying the determining conditions. Sections 1 and 2 were intended to be es-
sentially a review, but ultimately have some authorial input. Section 3 is half review and
half author due to the fact that the covariant Lorentz transformation (including indirectly
𝛽 𝛼 velocity) has been found in the literature, while the velocity 𝜀𝛼 have not been found in
the sources. And the next three sections of Part II deal with purely authorship issues as
planned. In addition to determining relative velocities, the article deals with the issue of
their addition in accordance with the article’s motto, which is a quote of prof. Zbigniew
Oziewicz.

1 The binary 4D relative velocity of Oziewicz–Świerk–Bolós and

its transitive-type addition with ternary properties

If the conditions v. and vi. are set for the binary velocity (1), then we can easily obtain the
canonical binary velocity of Oziewicz–Świerk–Bolós (see [26,27,36,4,17] and Fig.1):

𝑐2 𝛼
𝜔𝛼 (𝜈, 𝜇) = 𝜇 − 𝜈 𝛼 =∶ (𝜇 ⩤ 𝜈)𝛼 . (4)
𝜈◦𝜇

Condition vi. could be replaced by condition vii. in the form 𝜆1 = 1. On the other hand,
the 𝜆2 coefficient is equal to the reciprocal of the invariant Lorentz–Oziewicz gamma
coefficient for the binary velocity [18]:
𝜈◦𝜇 1 1
𝛾12 = 𝛾21 = =√ = . (5)
𝑐2 1 + 𝜔◦𝜔∕𝑐 2 𝜆 2

As it is easy to see the coefficients 𝛾12 , 𝜆2 are dependent on the scalar square of the
space-like binary velocity 𝜔◦𝜔 < 0.
It turns out that the conditions i. and ii. are not fulfilled √
here, because for parallel
velocities in the sense of 3D, the binary velocity is 𝛾𝑣 = 1∕ 1 − 𝑣2 ∕𝑐 2 times greater
(wherein 𝜈 𝑖 = 𝛾𝑣 𝑣𝑖 ). This means that canonical binary velocity can be written as [17]:

𝜔𝛼 = (𝜔0 , 𝜔) ⃗ ⋅ 𝑣∕𝑐,
⃗ = (𝛾𝑣 𝑤 ⃗
⃗ 𝛾𝑣 𝑤), (6)

where 𝑤 ⃗ = 𝜔∕𝛾
⃗ 𝑣 is 3D binary velocity [17] or jet velocity [18]. The 𝑤 ⃗ velocity satisfies
the full version of condition i. (not the limited condition ii.). In Part II of the work the
4D generalization of pseudo-binary velocity 𝑤 ⃗ will be given, called the cross velocity
𝜛𝛼 .
Binary velocity in the form of a 4D vector (4) appeared in 1988 in Świerk’s master
thesis [36], the originator and promoter of which was Oziewicz. Nevertheless Oziewicz
first published the canonical binary velocity only in 2004 [26] (perhaps it was blocked by
6 Grzegorz M. Koczan, GOL 2021 conference in memoriam Z. Oziewicz, Mexico

Fig. 1. Simple Bini–Carini–Jantzen’s geometric construction [3,20] of 4D relative velocity

𝜔𝛼12 (𝜈, 𝜇) called Oziewicz–Świerk–Bolós binary velocity [26,27,36,4,17], along with the inverse
binary velocity 𝜔𝛼21 (𝜇, 𝜈), in the Minkowski space-time diagram. The constructed relative veloci-
ties of two bodies with absolute Minkowski velocities 𝜈 𝛼 and 𝜇 𝛼 are determined by the appropriate
orthogonal line segments to these Minkowski velocities. Both binary velocities have equal values,
but indicate different directions in space-time.

reviewers [32]). A year later, the same velocity was published by Bolós [4], but for Gen-
eral Relativity. In a similar context, binary velocity (without a specific name) appears
in the newer work of [15]. On the other hand, in older works, such as Hestenes [14]
from 1974, relative velocity appears in the form of a bivector, not a vector. As a proof
of the fact that the important velocity vector 𝜔𝛼 is still not widely known, let the article
by a recognized relativist [1]. This article comes close to canonical binary velocity but
does not define its proper normalization. Despite the failure to fulfill the condition i. the
space-time normalization is correct and is equal to the norm for Einstein composition
𝜔𝛼 𝜔𝛼 = 𝜀𝛼 𝜀𝛼 = −|| ⃗ 2 (see Statement 2 in [17]). The author also came (by chance)
to binary velocity in 2019/2021 in the relativistic equation of motion of a body with a
variable own mass [18]. The idea of binary velocity apparently also appears in the 1995
work by Bini, Carini and Jantzen [3]. A distractor to find this work was to refer to the
General Relativity (also [4,15]), when the essence of the problem belongs to the Spe-
cial Relativity. In this context, binary velocity was introduced two years earlier (1993)
by Matolcsi [24], which he also described in his next work with Goher [25]. Binary
velocity also appears in the newer [8] work.
The binary relative velocity makes it possible to express the Minkowski velocity of
the second material point in terms of the Minkowski velocity of the first point:

𝜔𝛼 + 𝜈 𝛼
𝜇 𝛼 (𝜔, 𝜈) = √ =∶ (𝜔 ⟨+|𝜈)𝛼 . (7)
1 + 𝜔◦𝜔∕𝑐 2

This equation results from a simple transform (4) and the resulting square 𝜔◦𝜔 =
𝑐 6 ∕(𝜎◦𝜈)2 − 𝑐 2 . The operations ⩤ (4) and ⟨+| (7) are not internal operations. The op-
eration ⩤ is specified on Minkowski time-like velocities, but its value is a space-like
binary velocity. And ⟨+| works between binary and Minkowski velocity, and the value
is Minkowski velocity.
 Relative binary and ternary 4D velocities in SR: I. Covariant Lorentz transformation 7

Now consider composition the binary velocities themselves. Some kind of transitiv-
ity can be used in the definition of such composition:
( )𝛼
𝜔10 (𝜈, 𝜎) ⊞𝜎𝐵 𝜔02 (𝜎, 𝜇) ∶= 𝜔𝛼12 (𝜈, 𝜇), (8)
where: 𝜎 = {𝜎 𝛼 } – the Minkowski velocity of intermediate (third) material point;
10, 01, 12 – subscripts distinguish between three binary velocities and indicate a ref-
erence relation to pairs of material points (analogous to Fig. 1). The designation of the
addition of binary velocities ⊞𝜎𝐵 suggests its dependence on an additional parameter,
which was select Minkowski velocity of the material middle point. Equivalently, a differ-
ent point could be selected or the inverse velocity 𝜔−1
10
(𝜈, 𝜎) ∶= 𝜔01 (𝜎, 𝜈) ≠ −𝜔10 (𝜈, 𝜎)
alternately (compare with Fig. 1). Oziewicz used the latter choice.
Theorem 1. Composition (adding) two binary velocities 𝜔𝛼10 and 𝜔𝛼02 , meeting the con-
ditions of composition 𝜎◦𝜔10 ∕𝑐 2 = 𝛾10−1 − 𝛾 and 𝜎◦𝜔
10 02 = 0, is a ternary expression
depending on these two velocities and on the Minkowski velocity 𝜎 𝛼 of the middle ma-
terial point and is given by the formula:
𝜔𝛼02 + 𝜎 𝛼 𝜎𝛼
(𝜔10 ⊞𝜎𝐵 𝜔02 )𝛼 ∶= 𝜔𝛼10 + − , (9)
𝛾10 − 𝜔10 ◦𝜔02 ∕𝑐 2 𝛾10
√
where the invariant Lorentz–Oziewicz factor has the form 𝛾10 = 1∕ 1 + 𝜔10 ◦𝜔10 ∕𝑐 2 =
𝛾01 .
Proof. Using the relation (7) we calculate the Minkowski velocity 𝜇 𝛼 of the third mate-
rial point (No. 2) adding to the relative binary velocity 𝜔𝜇02 Minkowski velocity 𝜎 𝛼 :

𝜔𝛼02 + 𝜎 𝛼
𝜇𝛼 = (𝜔02 ⟨+| 𝜎)𝛼 = √ = 𝛾02 (𝜔𝛼02 + 𝜎 𝛼 ). (10)
1 + 𝜔02 ◦𝜔02 ∕𝑐 2
From the transformation of the analogous relation for the first (No. 1) and the second
material point (No. 0), it is possible to calculate the Minkowski velocity 𝜈 𝛼 of the first
material point:
1 𝛼
𝜈𝛼 = 𝜎 − 𝜔𝛼10 =∶ (𝜎 |−⟩ 𝜔10 )𝛼 , (11)
𝛾10
where, by the way, the subtraction operation of the inverse (in relation to 𝜎 𝛼 ) binary
velocity 𝜔10 (𝜈, 𝜎) = 𝜔−1
01
(𝜎, 𝜈) was defined. The scalar product of the first (No. 1) and
last Minkowski (No. 2) velocities is:
−1 2
𝜈◦𝜇 = 𝛾02 (𝛾10 (𝑐 + 𝜎◦𝜔02 ) − 𝜎◦𝜔10 − 𝜔10 ◦𝜔02 ) = 𝛾02 (𝛾10 𝑐 2 − 𝜔10 ◦𝜔02 ), (12)
where the assumptions (conditions of composition) were used. Thus, the composition
of the velocities, which is the binary velocity 𝜔𝛼12 takes the form:
𝜎 𝛼 + 𝜔𝛼02 𝜎𝛼
(𝜔10 ⊞𝜎𝐵 𝜔02 )𝛼 = 𝜔𝛼12 = (𝜇 ⩤ 𝜈)𝛼 = − + 𝜔𝛼10 , (13)
𝛾10 − 𝜔10 ◦𝜔02 ∕𝑐 2 𝛾10
which, with the accuracy of the order of writing, constitutes the thesis of the theorem.
⊔
⊓
8 Grzegorz M. Koczan, GOL 2021 conference in memoriam Z. Oziewicz, Mexico

To check compliance with Oziewicz’s original formula, we can use the useful equation
𝜔10 ◦𝜔02 = −𝛾10 𝜔−1 10
◦𝜔02 – for which the result value must be additionally changed
(remove the minus) due to other signatures (+ − −−) vs (− + ++).
Due to the fact that the composition of binary velocities depends on the middle mate-
rial point, and not the extreme point or inverse binary velocity, it is possible to refine the
associativity of velocities composition in the sense of Oziewicz, based on transitivity:
( ) ( )
𝜔10 (𝜈, 𝜎) ⊞𝜎𝐵 𝜔02 (𝜎, 𝜇) ⊞𝜇𝐵 𝜔23 (𝜇, 𝛽) = 𝜔10 (𝜈, 𝜎) ⊞𝜎𝐵 𝜔02 (𝜎, 𝜇) ⊞𝜇𝐵 𝜔23 (𝜇, 𝛽) ,
(14)
where both sides are by definition equal 𝜔13 (𝜈, 𝛽) = {𝜔𝛼13 (𝜈, 𝛽)}. The limited sense
of such associativity consists in numerous constraints (conditions) for composition of
binary velocities and in the ternary nature of such composition (the dependence on the
third parameter 𝜎 𝛼 or 𝜇𝛼 ).

2 The ternary Oziewicz–Ungar–Dragan relative velocity and

special case of its addition in the 4D Oziewicz–Einstein sense
Already composition of binary velocities shows ternarity and dependence on the dis-
tinguished frame of reference. Additionally, the binary composition required uncom-
fortable orthogonality conditions: the velocity 𝜔𝛼02 is othogonal to 𝜎 𝛼 , but velocities
𝜔𝛼10 , 𝜔𝛼12 are ortgononal to 𝜈 𝛼 . It is therefore worth considering a velocities subset that
will contain orthogonal velocities up to the one velocity only (e.g. to 𝜎 𝛼 – see Fig. 2).
Simply put, Oziewicz followed this path, introducing a relative ternary velocity [26,28].
Above all, however, he referred to the Lorentz transformation (in the active sense iii.,
not resting sense iv.).
In this work, the ternary velocity will be introduced directly at the level of the co-
variant Lorentz transformation described in the next section. It should be known that
the Lorentz bost depends on the bivector, which can be described by two Minkowski ve-
locities (reference velocity 𝜎 𝛼 and bost velocity 𝜁 𝛼 ) or equivalently Minkowski velocity
(𝜎 𝛼 ) and Oziewicz velocity (𝜔𝛼 or 𝜉 𝛼 ) orthogonal to the first (𝜔◦𝜎 = 0 or 𝜉◦𝜎 = 0):

(𝜎 ∧ 𝜁 )𝛼𝛽 ∶= 𝜎 𝛼 𝜁 𝛽 − 𝜎 𝛽 𝜁 𝛼 = 𝜎 𝛼 𝜉 𝛽 − 𝜎 𝛽 𝜉 𝛼 =∶ (𝜎 ∧ 𝜉)𝛼𝛽 . (15)

A more natural parameterization seems to be Oziewicz space-like velocity (ternary 𝜉 𝛼

or binary 𝜔𝛼 ), as it is normalized (with the signature minus) to the usual boost velocity
value.
Nevertheless, we will begin by defining a Minkowski velocity 𝜁 𝛼 that parameterizes
the following active Lorentz boost (or equivalent passive inverse boost):
[ ] [ ]
𝐿𝑎(𝜎, 𝜁) {𝜈 𝛼 } ∶= {𝜇𝛼 } ⟷ 𝐿𝑝(𝜎, 𝜁) {𝜇𝛼 } ∶= {𝜈 𝛼 }. (16)

Thanks to the Theorem 7 proved in the next section (see (70)), it is possible to provide
a solution to the above equation taking into account the notations used here:

𝜎◦𝜈 + 𝜎◦𝜇 𝑐 2 − 𝜈◦𝜇 − 2(𝜎◦𝜈)2 ∕𝑐 2

𝜁𝛼 = (𝜇𝛼 − 𝜈 𝛼 ) − 𝜎 𝛼 . (17)
𝑐 2 − 𝜈◦𝜇 + 2(𝜎◦𝜈)(𝜎◦𝜇)∕𝑐 2 𝑐 2 − 𝜈◦𝜇 + 2(𝜎◦𝜈)(𝜎◦𝜇)∕𝑐 2
 Relative binary and ternary 4D velocities in SR: I. Covariant Lorentz transformation 9

𝛼
Fig. 2. Visualization of the Oziewicz–Ungar–Dragan ternary relative velocity 𝜉(𝜎) and the
𝜇 𝛼
Einstein–Oziewicz relative velocity 𝜀12 (𝜎, 𝜈, 𝜇), along with the inverse velocities −𝜉(𝜎) and
𝜇
𝜀21 (𝜎, 𝜇, 𝜈), on the space-time Minkowski diagram . Apart from the Minkowski absolute velocities
𝜈 𝛼 and 𝜇 𝛼 of two material points, the above relative velocities depend on the Minkowski velocity
𝜎 𝛼 of the observer, what illustrates the relativistic property of ternarity. The Oziewicz–Ungar–
𝛼
Dragan ternary velocity 𝜉(𝜎) is a space-like parameter (inverse velocity in another sense) of the
active Lorentz boost transforming of 𝜈 𝛼 to 𝜇𝛼 . Whereas, the Einstein–Oziewicz velocity 𝜀𝜇12 is a
space-like version of the velocity transformation 𝜇𝛼 of passive Lorentz boost to the system with
the velocity 𝜈 𝛼 .

Based on the scalar product:

(𝜎◦𝜈)2 + (𝜎◦𝜇)2 + 𝑐 2 𝜈◦𝜇 − 𝑐 4

𝜁 ◦𝜎 = , (18)
𝑐 2 − 𝜈◦𝜇 + 2(𝜎◦𝜈)(𝜎◦𝜇)∕𝑐 2
𝛼 = (𝜁 ⩤
it is already possible to determine the canonical ternary Oziewicz velocity: 𝜉(𝜎)
𝜎)𝛼 as relative velocity (in binary sense) 𝜁 𝛼 (𝜎, 𝜈, 𝜇) with respect to 𝜎 𝛼 :

𝛼 𝑐 2 𝜎◦𝜈 + 𝑐 2 𝜎◦𝜇 (𝜎◦𝜇)2 − (𝜎◦𝜈)2

𝜉(𝜎) = (𝜇𝛼 − 𝜈 𝛼 ) − 𝜎𝛼 .
(𝜎◦𝜈)2 + (𝜎◦𝜇)2 + 𝑐 2 𝜈◦𝜇 − 𝑐 4 (𝜎◦𝜈)2 + (𝜎◦𝜇)2 + 𝑐 2 𝜈◦𝜇 − 𝑐 4
(19)
This could be written a little shorter (see [17,18]):

𝑐 2 𝜎◦𝜈 + 𝑐 2 𝜎◦𝜇 ( 𝜎◦𝜇 − 𝜎◦𝜈 𝛼 )

𝛼
𝜉(𝜎) (𝜎, 𝜈, 𝜇) = 𝜇𝛼 − 𝜈 𝛼 − 𝜎 . (20)
(𝜎◦𝜈)2 + (𝜎◦𝜇)2 + 𝑐 2 𝜈◦𝜇 − 𝑐 4 𝑐2
This velocity in this work is called Oziewicz–Ungar–Dragan ternary velocity (Fig. 2),
according Oziewicz [26,27,28], even though Ungar [40,41] and Dragan [10,11] only
gave its 3D version (see simple interpretation by the laufer/bishop method [19]), which
is not explicitly ternary. Moreover, the Ungar approach was not about relative velocity
(subtraction), but addition ("Einstein coaddition"). In order to expose this ternarity, one
can introduce the ternary subtraction of Minkowski velocities leading to the ternary
velocity:
(𝜇 ⨺𝜎 𝜈)𝛼 ∶= 𝜉(𝜎)𝛼
(𝜎, 𝜈, 𝜇). (21)
10 Grzegorz M. Koczan, GOL 2021 conference in memoriam Z. Oziewicz, Mexico

𝛼 :
This does not contradict binary relationships between velocities 𝜁 𝛼 and 𝜉(𝜎)

𝛼
𝜉(𝜎) = (𝜁 ⩤ 𝜎)𝛼 = 𝜔𝛼 (𝜎, 𝜁) ⟷ 𝜁 𝛼 = (𝜉(𝜎) ⟨+| 𝜎)𝛼 . (22)

On the other hand, the more direct relations (without 𝜁 𝛼 ) are ternary (see Fig. 2):
𝛼
𝜉(𝜎) = (𝜇 ⨺𝜎 𝜈)𝛼 ⟷ 𝜇𝛼 =∶ (𝜉(𝜎) ⨹𝜎 𝜈)𝛼 . (23)

Unfortunately, all of the above operations are not internal operations. As for binary
velocities, also for ternary velocities there should be a certain rule of their addition. In
Part II of the work this problem will be attacked in general, and now Oziewicz interpre-
tation related to Einstein velocity composition will be given.The key here is the relation
of the Lorentz transformation in the form (16). It means that the "orthodox" compo-
sition of Minkowski velocities 𝜁 𝛼 and 𝜈 𝛼 is Minkowski velocity 𝜇𝛼 . Now it is enough
to convert the time-like Minkowski velocities into relative space-like velocities (binary
and ternary Oziewicz velocities). So, if we define the addition of ternary velocities in
the sense of Oziewicz as follows:
( )𝛼
𝜉(𝜎)01 (𝜎, 𝜎, 𝜈) ⊞𝜎𝑂 𝜉(𝜎)12 (𝜎, 𝜈, 𝜇) ∶= 𝜉(𝜎)02
𝛼
(𝜎, 𝜎, 𝜇) = 𝜔𝛼02 (𝜎, 𝜇), (24)

it will be, in a sense, Lorentz boost:

( ( ))𝛼
(𝜈 ⩤ 𝜎) ⊞𝜎𝑂 𝜁 (𝜈 → 𝜇) ⩤ 𝜎 ∶= (𝜇 ⩤ 𝜎)𝛼 . (25)

The sense of such Lorentz boost is developed in the following theorem. It is worth noting
that ternary addition in the sense of Oziewicz cannot be associative, because one of the
added velocities (as well as the resulting velocity) is ordinary binary velocity, not ternary
(it depends twice on 𝜎).

Theorem 2 (Oziewicz’s). Ternary velocities composition in the sense of Oziewicz ⊞𝜎𝑂

(adding ternary velocity 𝜉 𝛼 ≡ 𝜉(𝜎) 𝛼 ) in the material point system 𝜎 𝛼 is the 4D version of

Einstein’s 3D velocity addition (⊕0 ) and has an analogous algebraic form:

𝛾𝜉 ( )𝛼 𝛾𝜉
( )
𝜉◦𝜉 𝛼 𝜔◦𝜉 𝛼
𝜔𝛼 + 𝜉 𝛼 + 𝛾 +1 (𝜔 ∧ 𝜉) ⊙ 𝜉 ∕𝑐 2 𝜔𝛼 + 𝜉 𝛼 + 𝛾 +1 𝑐 2 𝜔 − 𝑐 2 𝜉
𝜉 𝜉
(𝜔 ⊞𝜎𝑂 𝜉)𝛼 = = .
1 − 𝜔◦𝜉∕𝑐 2 1 − 𝜔◦𝜉∕𝑐 2
(26)
In order to refer to the 3D version, it is enough to convert tensor products (antisymmetric
and constricted) to 3D vector products and to change to opposite signs of scalar products
(transition from signature + − −− to signature + + +):
( ⃗ )
⃗ + 𝛾𝑊 (𝑣⃗ × 𝑊 ⃗ )×𝑊 ⃗ ∕𝑐 2 ⃗ + 𝛾𝑊 𝑣⋅ ⃗𝑊 ⃗ 𝑊2
𝑣⃗ + 𝑊 𝛾𝑊 +1
𝑣
⃗ + 𝑊 𝛾𝑊 +1 𝑐 2 𝑊 − 𝑐 2 𝑣
⃗
𝑣⃗ ⊕0 𝑊⃗ = = .
⃗
1 + 𝑣⃗ ⋅ 𝑊 ∕𝑐 2 ⃗
1 + 𝑣⃗ ⋅ 𝑊 ∕𝑐 2
(27)
Thus, the addition of the ternary velocity in the sense of Oziewicz can rightly be called
the addition of Oziewicz–Einstein.
 Relative binary and ternary 4D velocities in SR: I. Covariant Lorentz transformation 11

Proof. In the 3D Dragan approach, the relationship between the ternary velocity (as "in-
verse" velocity) with the Lorentz transformation and Einstein velocities composition is
not in doubt and results almost from the very definition of ternary velocity (see [17,19]).
Oziewicz 4D approach is more general and, apart from the definition of ternary velocity,
must be based on the definition of its addition in the form (24) or (25). This makes the
examined thesis a non-obvious theorem interpreting Einstein velocities composition in
terms of ternary velocity – since it states that 4D formulas are fully analogous to 3D
formulas. The latter statement requires a formal proof by calculations.
In order to use the definition (25) we first apply the covariant Lorentz transformation
in the active version (see (61)):
𝜈◦(𝜁 + 𝜎) 𝛼 𝜈◦𝜎 𝛼
𝜇𝛼 = 𝐿𝑎𝛼𝛽 (𝜎, 𝜁) 𝜈 𝛽 = 𝜈 𝛼 − (𝜁 + 𝜎 𝛼 ) + 2 𝜁 . (28)
𝜎◦(𝜁 + 𝜎) 𝑐2
The needed scalar products can be expressed as follows:

𝜈◦𝜎 = 𝛾𝜈 𝑐 2 = 𝛾𝜔 𝑐 2 , 𝜁◦𝜎 = 𝛾𝜁 𝑐 2 = 𝛾𝜉 𝑐 2 , (29)

( )
𝜈◦𝜁 = 𝛾𝜔 𝛾𝜉 𝑐 2 − 𝑣⃗ ⋅ 𝑊
⃗ = 𝛾𝜔 𝛾𝜉 (𝑐 2 + 𝜔◦𝜉), (30)
𝜈◦𝜎
𝜇◦𝜎 = 2 𝜁 ◦𝜎 − 𝜈◦𝜁 = 𝛾𝜔 𝛾𝜉 (𝑐 2 − 𝜔◦𝜉). (31)
𝑐2
Now it is enough to go from Minkowski velocities to space-like (binary or ternary)
velocities: 𝜇 → 𝜇 ⩤ 𝜎, 𝜈 = 𝜔⟨+|𝜎, 𝜁 = 𝜉⟨+|𝜎:
( )
𝜔𝛼 + 𝜎 𝛼 𝛾𝜉 + 1 + 𝛾𝜉 𝜔◦𝜉
𝑐2
(𝛾𝜉 𝜉 𝛼 + 𝛾𝜉 𝜎 𝛼 + 𝜎 𝛼 ) 𝜉 𝛼 + 𝜎 𝛼
𝛼
(𝜇⩤𝜎) = ( 𝜔◦𝜉 )
− ( 𝜔◦𝜉 )
+2 𝜔◦𝜉
−𝜎 𝛼 , (32)
𝛾𝜉 1 − 𝑐 2 𝛾𝜉 1 − 𝑐 2 (𝛾𝜉 + 1) 1 − 𝑐2

which, after algebraic transformations, leads to the thesis of theorem. ⊔

⊓

It is worth noting that Oziewicz-Einstein addition of the ternary velocity does not
depend explicitly on 𝜎 𝛼 , but only implicitly in the form of v. type assumptions about
orthogonality: 𝜔◦𝜎 = 𝜉◦𝜎 = (𝜔 ⊞𝜎𝑂 𝜉)◦𝜎 = 0. It should not be thought that Oziewicz’s
theorem only indicates the change of the language of Minkowski’s time-like velocities
into space-like (ternary and binary) velocities. First, the ternary velocity itself is some-
thing significantly different from Einstein’s composition of velocities. This is evidenced
by the fulfillment of assumption iii. in opposition to the assumption iv. – and fulfilling
ii. in opposition to the assumption i. Moreover, ternary relative velocity is antisymmet-
ric (assumption viii.), and Einstein composition is not antisymmetric in subtracting, nor
symmetrical in adding. Secondly, adding ternary velocity in the sense of Oziewicz–
Einstein is a special case of ternary composition – the general case will be attacked in
Part II of the paper.
The concept of relative ternary velocity was published in full general by Oziewicz in
2004 [26] and 2007 [27] (see also [28,29,31]). A 3D version of this velocity for addition
was published in the Ungar handbook in 2001 [40] (see also [41]). Definitions of this
velocity were published by Ungar as early as 1992, but at that time he did not provide the
result of its calculation [39]. However, Ungar did not recognize the special importance
12 Grzegorz M. Koczan, GOL 2021 conference in memoriam Z. Oziewicz, Mexico

of ternary velocity ("Einstein coaddition") and preferred other velocities as well: proper
velocity [42], Möbius velocity addition, Möbius coaddition [43]. The author has already
quoted in [18] the velocity given by Ungar [43], but in a differentially equivalent 3D
ternary velocity version, because Ungar did not provide the full formula for this velocity
in this paper.
The Lorentz transformation in terms of initial and final vectors developed along a
parallel path (Wyk 1986, 1991 [45,46], also Ungar 1992 [39]). This concept is similar to
the covariant Lorentz transformation, discussed later. A slightly similar approach based
on the Lorentz transformation of the 4D position vector was used by Urbantke in 2002
[44]. The approaches of these three researchers (Wyk, Ungar, Urbantke), however, were
less general than that of Oziewicz (based on [5,6]). The initial and final vectors approach
(Lorentz link boost problem) was described by Celakoska in 2008 [5] and Celakoska and
Chakmakov in 2011 [6]. In 2015, however, Celakoska’s team published a quite clear for-
mulae for the relative 4D and 3D ternary velocities [8]. The Lorentz link boost approach
was used, but the 3D result was similar to the Urbantke formula. Unfortunately, for the
subtracted 3D velocities, a different normalization was used (like the Minkowski four-
velocity), which for the resulting 3D ternary velocity – it simplified the formula, but
made it difficult to read it correctly.
On the other hand, a strictly 3D approach was used in 2012 by Dragan in resuming
his lectures [10]. The limitation to 3D description makes it impossible to full recognize
the ternary feature of relative velocity. Nevertheless, Dragan presented a slightly more
general approach than Urbantke ("magic four-rule") in the further chapter of his lectures
[10,11] (see also [30,31]).
The use of Oziewicz–Ungar–Dragan ternary velocity allowed the author to define a
new concept of ternary relativistic acceleration [18,17,19,20]. This acceleration solved
the problem of the non-parallelism of the force to acceleration in three-dimensional
space in Special Relativity. This, in turn, made it possible to define the inertial mass
unifying the transverse and longitudinal mass. The use of the same method of defining
acceleration for Einstein velocity composition, instead of Oziewicz–Ungar–Dragan ve-
locity, does not solve these problems of relativistic dynamics – see the work of Celakoska
and Chakmakov [7]. Celakoska and Chakmakov did not use the relative ternary veloc-
ity (Oziewicz-Ungar-Dragan), probably because they did not recognize the fundamental
role of this velocity, which they themselves described earlier [8].
The discovery of the ternarity of relative velocity in a way reflects Oziewicz’s anti-
Popperian philosophy of science. Contrary to Popper’s postulates, Oziewicz draws atten-
tion to the need for individual and subjective scientific research conducted by specialists
[33]. The dependence of the relative velocity of two bodies on the observer is a specific
example of this philosophy of anti-objectivism. A subjectively preferred frame of refer-
ence [9] may be associated with some idea of Æther which is occasionally considered
[30,35]. It should be emphasized, however, that ternary velocity is an orthodox concept
in Special Relativity. We find some resemblance to anti-Popperian philosophy of sci-
ence in the Pythagoreans, who did not apply the principles of democracy to science.
However, this should be understood as freedom of research, not as monopoly on it by
any group [32].
 Relative binary and ternary 4D velocities in SR: I. Covariant Lorentz transformation 13

2.1 3D formulae of relative ternary velocity

While the article is essentially written in 4D language, some things may seem (and
are) simpler in terms of 3D. In the case of ternary velocity, the simplification of the
3D formula may result from the apparent loss of ternary property, which is hidden in
the selection of the 3D reference frame. In any case, it’s worth knowing the explicit
3D formulas of Ungar, Dragan, Urbantke and Celakoska’s team in order to be able to
directly compare them with Einstein’s formula.
Let the Cartesian frame fit to the velocity of the Minkowski frame of reference
𝛼 ⃗ Then the relative ternary velocity of the body with the velocity of Minkowski
𝜎 = (𝑐, 0).
𝜇 = (𝛾𝑢 𝑐, 𝛾𝑢 𝑢⃗) in relation to the body with the velocity of Minkowski 𝜈 𝛼 = (𝛾𝑣 𝑐, 𝛾𝑣 𝑣)
𝛼 ⃗
is practically reduced to the three-dimensional velocity 𝜉 𝛼 = (0, 𝑊 ⃗ ). The 3D ternary
velocity 𝑊 ⃗ can be expressed in three forms differing in the way of writing the normal-
ization coefficient. Here is the formula given by Ungar ("Einstein coaddition") [40,41],
changed to subtraction :
𝛾𝑢 + 𝛾𝑣
⃗ = 𝑢⃗ ⊟ 𝑣⃗ ∶=
𝑊 (𝛾𝑢 𝑢⃗ − 𝛾𝑣 𝑣),
⃗ (33)
⃗ 2 ) + 𝛾𝑢2 + 𝛾𝑣2 − 1
𝛾𝑢 𝛾𝑣 (1 − 𝑢⃗ ⋅ 𝑣∕𝑐

which is most compatible with the 4D Oziewicz formula [17]. Dragan’s original formula
is slightly different [10,11,18]:

𝛾𝑢−1 + 𝛾𝑣−1
⃗ = 𝑢⃗ ⊖∧ 𝑣⃗ ∶=
𝑊 (𝛾𝑢 𝑢⃗ − 𝛾𝑣 𝑣).
⃗ (34)
⃗ 2 + 𝛾𝑢 𝛾𝑣 (1 − 𝑢2 𝑣2 ∕𝑐 4 )
1 − 𝑢⃗ ⋅ 𝑣∕𝑐

However, the simplest is the formula resulting from the work Urbantke [44] written for
relative velocity by Celakoska’s team [8] in the form:

2(𝛾𝑢 + 𝛾𝑣 )
⃗ = 𝑢⃗ ⊟∧ 𝑣⃗ ∶=
𝑊 (𝛾𝑢 𝑢⃗ − 𝛾𝑣 𝑣).
⃗ (35)
(𝛾𝑢 + 𝛾𝑣 )2 ⃗ 2 ∕𝑐 2
+ (𝛾𝑢 𝑢⃗ − 𝛾𝑣 𝑣)

The above equation will come out if we change two four-positions into two four-velocities
in the Urbantke velocity formula (see further (44)) – but Urbantke did not do it in [44].
Oziewicz referred to the original formula of Urbantke [30,31]. Ungar also defined a for-
mula of the Urbantke type, but did not write it out full explicitly [40]. As already men-
tioned, a slightly more general formula than Urbantke was proposed by Dragan, which
he called "magic four-rule" [10,11]. However, strictly speaking, this is not a strictly 4D
rule, nor strictly 3D, but it describes the correspondence between 4D and 3D (and vice
versa).

Theorem 3 (About 3D ternary velocity formulae). The formulae of Ungar (33), Dra-
gan (34) and Urbantke-Celakoska (35) of the ternary velocity 𝑊 ⃗ are identity equal
(⊟ ≡ ⊖∧ ≡ ⊟∧ ). These formulae for ternary relative velocity 𝑊 ⃗ express the veloc-
ity of the active Loretnz boost "accelerating" the velocity 𝑣⃗ to velocity 𝑢⃗: 𝐿𝑎𝑊⃗ (𝑣)
⃗ = 𝑢⃗,
or express the velocity of the passive Lorentz boost "delaying" the velocity of 𝑢⃗ to the
⃗ 𝐿𝑝𝑊⃗ (⃗𝑢) = 𝑣.
velocity of 𝑣: ⃗
14 Grzegorz M. Koczan, GOL 2021 conference in memoriam Z. Oziewicz, Mexico

Proof. We’ll start by transforming the denominator of formula (35):

⃗2
(𝛾𝑢 𝑢⃗ − 𝛾𝑣 𝑣) 𝑢2 𝑣2 𝑢⃗ ⋅ 𝑣⃗
(𝛾𝑢 + 𝛾𝑣 )2 + = 𝛾𝑢2 + 𝛾𝑣2 + 2𝛾𝑢 𝛾𝑣 + 𝛾𝑢2 + 𝛾𝑣2 − 2𝛾𝑢 𝛾𝑣
𝑐2 𝑐 2 𝑐2 𝑐2
𝑢⃗ ⋅ 𝑣⃗
= 𝛾𝑢2 + 𝛾𝑣2 + 2𝛾𝑢 𝛾𝑣 + 𝛾𝑢2 (1 − 𝛾𝑢−2 ) + 𝛾𝑣2 (1 − 𝛾𝑣−2 ) − 2𝛾𝑢 𝛾𝑣 (36)
𝑐2
( )
𝑢
⃗ ⋅ 𝑣
⃗
= 2𝛾𝑢2 + 2𝛾𝑣2 − 2 + 2𝛾𝑢 𝛾𝑣 1 − ,
𝑐2
which is equal to double the denominator of the formula (33). Since the numerators also
differ twice, the formulas (33) and (35) are indentity equal.
Now let’s transform the last term of the formula denominator of (34):

𝛾𝑢 𝛾𝑣 (1 − 𝑢2 𝑣2 ∕𝑐 4 ) = 𝛾𝑢 𝛾𝑣 − 𝛾𝑢 𝛾𝑣 (1 − 𝛾𝑢−2 )(1 − 𝛾𝑣−2 ) = 𝛾𝑢 𝛾𝑣−1 + 𝛾𝑢−1 𝛾𝑣 − 𝛾𝑢−1 𝛾𝑣−1 . (37)

If we take this into account and extend the numerator and denominator of the formula
(34) with the factor 𝛾𝑢 𝛾𝑣 , we get the formula (33). So the formulae (33),(34),(35) are
indentity equal.
Now consider the passive Lorenzt transformation, written for the direction parallel
and perpendicular to the direction of velocity 𝑊⃗:
( ) ( )
𝑟⃗′∥ = 𝛾𝑊 𝑟⃗∥ − 𝑊⃗ 𝑡 , 𝑟⃗′ = 𝑟⃗⟂ , 𝑡′ = 𝛾𝑊 𝑡 − 𝑊
⟂
⃗ ⋅ 𝑟⃗∕𝑐 2 . (38)

The spatial part of the transformation can be expressed by one equation:

2
𝛾𝑊 𝑟⃗ ⋅ 𝑊⃗
𝑟⃗′ = 𝑟⃗ − 𝛾𝑊 𝑊
⃗ 𝑡+ ⃗.
𝑊 (39)
2
𝛾𝑊 +1 𝑐 2

If the dot product in this equation is expressed in terms of times, then from the equation:
2
𝛾𝑊 ( )
′ ⃗ 𝑡+ 𝑡′ ⃗
𝑟⃗ = 𝑟⃗ − 𝛾𝑊 𝑊 𝑡− 𝑊, (40)
2 +1
𝛾𝑊 𝛾𝑊

the velocity can be easily calculated:

′
⃗ = (1 + 𝛾 −1 ) 𝑟⃗ − 𝑟⃗ .
𝑊 (41)
𝑊 𝑡 + 𝑡′

−2 :
We still have to compute 𝛾𝑊 . For this purpose, we use the relationship 𝑊 2 ∕𝑐 2 = 1−𝛾𝑊
( )2
−1 2 𝑟⃗ − 𝑟⃗′
(1 + 𝛾𝑊 ) ∕𝑐 2 = 1 − 𝛾𝑊
−2
. (42)
𝑡 + 𝑡′
The above equation after algebraic simplification becomes a linear equation for 𝛾𝑊 , the
solution of which is:
( −⃗𝑟′ )2 2
1 + 𝑟⃗𝑡+𝑡 ′ ∕𝑐 2
−1
𝛾𝑊 = ( 𝑟⃗−⃗𝑟′ )2 ⟷ 1 + 𝛾𝑊 = ( 𝑟⃗−⃗𝑟′ )2 . (43)
1 − 𝑡+𝑡′ ∕𝑐 2 1 + 𝑡+𝑡′ ∕𝑐 2
 Relative binary and ternary 4D velocities in SR: I. Covariant Lorentz transformation 15

Using this result, we obtain the Urbantke formula [44]:

⃗ = 2 𝑟⃗ − 𝑟⃗′
𝑊 ( 𝑟⃗−⃗𝑟′ )2 ′
. (44)
1+ 𝑡+𝑡′
∕𝑐 2 𝑡 + 𝑡

This formula expresses the Lorentz transformation velocity – the most fundamental rel-
ative velocity we can imagine – Einstein’s transformed relative velocity is a velocity
secondary to that (physically and mathematically, but not historically).
In the equation (44) the components of four-position vectors can be replaced by
components of any four-vector. So if we change the four-position (𝑐𝑡, 𝑟⃗) to the four-
velocity (𝛾𝑢 𝑐, 𝛾𝑢 𝑢⃗), and the four-position (𝑐𝑡′ , 𝑟⃗′ ) to the four-velocity (𝛾𝑣 𝑐, 𝛾𝑣 𝑣)
⃗ we get
the formula for the ternary velocity (35). Thus, in fact, this formula describes a passive
Lorenzt transformation 𝐿𝑝𝑊⃗ (⃗𝑢) = 𝑣. ⃗ And the active transformation obviously works in
the opposite direction 𝐿𝑎𝑊⃗ (𝑣) ⃗ = 𝑢⃗ which was the last thing to prove. ⊔
⊓

3 Explicitly covariant rewriting of Einstein composition of 4D

Minkowski velocities in the ternary sense

The most typical Lorentz transformation is the so-called Lorentz boost. This boost in
the 3D formalism is parameterized with three-dimensional velocity. In addition, the 3D
velocity itself is subject to other transformation laws resulting from the Lorentz trans-
formation (Einstein velocities composition). The Lorentz boost formula for coordinates
and the 3D velocities composition formula are, however, significantly different despite
their formal closeness. The consequence of this difference is the rotation of Tomas–
Wigner accompanying the composition of non-parallel 3D velocities, as well as the lack
of associativity for the composition velocities operation – although Lorentz boosts com-
position is associative by definition of the Lorentz group (however, it no longer has to
be considered a pure boost).
If the 3D velocities composition formula is written in 4D form, the above duality
(Lorentz boost of coordinates vs velocities composition) disappears in a way. This is
how the covariant Lorentz transformation (for the Lorentz boost case) will be derived
in this work. Of course, the covariant Lorentz transformation can be introduced differ-
ently, but the author succeeded precisely on the basis of the 4D generalization of Einstein
velocities composition. Only then was it possible to find an analogous formula for the
covariant Lorentz transformation in literature. Unfortunately, it is not widely used in the
literature and, apart from the source [37] and pioneering works from about half a century
ago [2,22,23,13], it is difficult to indicate many works containing the covariant Lorentz
transformation – let alone its usage. A distractor that hinders the recognition of the co-
variant Lorentz transformation is its ternary character (dependence on the Minkowski
velocity of the distinguished frame of reference 𝜎 𝛼 ). It was this distractor that disrupted
the author in finding the covariant Lorentz transformation while writing his master the-
sis [21]. However, the thesis [21] contained an almost covariant form of the Lorentz
transformation, and only the introduction of the 𝜎 𝛼 vector was missing (e.g. 𝜎◦𝑈 ∕𝑐 in
place of 𝑈0 ).
16 Grzegorz M. Koczan, GOL 2021 conference in memoriam Z. Oziewicz, Mexico

We’ll start by writing the full 3D Einstein velocities composition formula. Both in
the context of relative velocity and in the context of the passive Lorentz transformation,
such a law is naturally a subtraction rather than an addition of velocity (contrary to
Einstein’s original [12]):
𝛾𝑣
𝑢⃗ − 𝑣⃗ + 𝛾𝑣 +1
(⃗𝑢 × 𝑣) ⃗ 2
⃗ × 𝑣∕𝑐
′ ⃗ 𝑣,
𝑢⃗ ⊖0 𝑣⃗ ∶= 𝑢⃗ = =∶ ( ⃗ 𝑢⃗), (45)
⃗ 2
1 − 𝑢⃗ ⋅ 𝑣∕𝑐

where ⃗ is the relative velocity of a body moving at velocity 𝑢⃗ relative to a body moving
at velocity 𝑣.
⃗ Now we have to rewrite the above law using space-like 4D (binary) relative
velocities. Let 𝑣⃗ → 𝜈 𝜇 → 𝜔𝛼01 = (𝜈 ⩤ 𝜎)𝛼 and 𝑢⃗ → 𝜇𝛼 → 𝜔𝛼02 = (𝜇 ⩤ 𝜎)𝛼 . Based on
⃗ 𝑣⃗ − 𝑣2 𝑢⃗ the formula (45) can be written as follows:
⃗ × 𝑣⃗ = (⃗𝑢 ⋅ 𝑣)
the identity (⃗𝑢 × 𝑣)
( )
𝛾01 𝜔01 ◦𝜔01 𝛼 𝜔02 ◦𝜔01 𝛼
𝜔𝛼02 − 𝜔𝛼01 + 𝛾01 +1 𝑐2
𝜔02 − 𝑐2
𝜔01
(𝜔02 ⊟𝜎0 𝜔01 )𝛼 ∶= =∶ 𝜀𝛼 (𝜔01 , 𝜔02 ),
1 + 𝜔02 ◦𝜔01 ∕𝑐 2
(46)
where ⊟𝜎0 is a 4D generalization from the definition of a 3D operation ⊖0 . Due to the
signature selection, the signs before the 4D scalar products have been changed com-
pared to the 3D scalar products (⃗𝑢 ⋅ 𝑣⃗ → −𝜔02 ◦𝜔01 ).Thus, Einstein velocities composi-
tion is treated here as some composition of binary velocities – but other than transitive
Oziewicz composition of binary velocities. We already know from the previous section
that the Lorentz boost velocity 𝜔𝛼01 can be interpreted up to the sign as ternary velocity.
It can be concluded that the inverse implication to Oziewicz’s theorem is derived now.
Thus, the form of ⊟𝜎0 (and consequently ⊞𝜎0 ∶= −⊟𝜎0 ) is a definition, and Oziewicz’s
theorem was the thesis that ⊞𝜎𝑂 = ⊞𝜎0 . In other words, the Oziewicz operation ⊞𝜎𝑂
(Oziewicz–Eisntein operation) was defined differently than the usual 4D generalization
⊞𝜎0 of 3D Einstein operation ⊕0 ∶= −⊖0 . In any case, using the action ⊟𝜎𝑜 we ob-
tained the space-like velocity 𝜀𝛼 , which will be called the relative velocity of Einstein–
Oziewicz. The Einstein–Oziewicz velocity 𝜀𝛼 depends explicitly only on two space-like
composition (subtracted) velocities, which are orthogonal to the Minkowski velocity 𝜎 𝛼
of reference frame.
Now we will express the Einstein–Oziewicz relative velocity 𝜀𝛼 using the Minkowski
velocities 𝜈 𝛼 , 𝜇𝛼 , as well as 𝜎 𝛼 , which will directly reflect the ternary character of the
Einstein–Oziewicz relative velocity (see Fig. 2).

Theorem 4. The space-like Einstein-Oziewicz relative velocity 𝜀𝛼 of the material point

𝜇𝛼 in relation to the material point 𝜈 𝛼 in the reference frame of velocity 𝜎 𝛼 is expressed
as follows in terms of the three above-mentioned Minkowski velocities:
[ ]
𝑐2 𝜇◦(𝜈 + 𝜎) 𝛼
𝜀𝛼 (𝜎, 𝜈, 𝜇) = 𝜇𝛼 − (𝜈 + 𝜎 𝛼 ) + 𝜎 𝛼 =∶ (𝜇[−]𝜎0 𝜈)𝛼 , (47)
𝜇◦𝜈 𝜎◦(𝜈 + 𝜎)

which also specifies a new type (ternary or pseudo-binary) of Minkowski velocities sub-
traction [−]𝜎0 with a space-like value.
 Relative binary and ternary 4D velocities in SR: I. Covariant Lorentz transformation 17

Proof. Binary velocities appearing in the formula (46) have to be written into the Minkowski
velocities 𝜔𝛼01 = (𝜈 ⩤ 𝜎)𝛼 and 𝜔𝛼02 = (𝜇 ⩤ 𝜎)𝛼 :
( ))
𝑐2 𝛼 𝑐2 𝛼 𝜈◦𝜎 𝜔01 ◦𝜔01 ( 𝑐 2 𝛼 ) 𝜔02 ◦𝜔01 ( 𝑐 2 𝛼
𝜇◦𝜎
𝜇 − 𝜈◦𝜎
𝜈 + 𝜈◦𝜎+𝑐 2 𝑐2 𝜇◦𝜎
𝜇 − 𝜎𝛼 − 𝑐2 𝜈◦𝜎
𝜈 − 𝜎𝛼
𝜀𝛼 = .
1 + 𝜔02 ◦𝜔01 ∕𝑐 2
(48)
And the scalar products are equal to:

𝑐6
𝜔01 ◦𝜔01 = −𝑣2 = − 𝑐2, (49)
(𝜈◦𝜎)2

(𝜇◦𝜈)𝑐 4
𝜔02 ◦𝜔01 = − 𝑐2. (50)
(𝜇◦𝜎)(𝜈◦𝜎)
We can simplify the computational complexity by introducing auxiliary symbols: 𝑥 ∶=
𝜈◦𝜎∕𝑐 2 , 𝑦 ∶= 𝜇◦𝜎∕𝑐 2 , 𝑧 ∶= 𝜇◦𝜈∕𝑐 2 . Then:
( 2( ) ( ))
1 𝛼 1 𝛼 𝑥 1−𝑥 1 𝛼 𝛼 − 𝑧−𝑥𝑦 1 𝜈 𝛼 − 𝜎 𝛼
𝑦
𝜇 − 𝑥
𝜈 + 𝑥+1 𝑥2 𝑦
𝜇 − 𝜎 𝑥𝑦 𝑥
𝜀𝛼 = 𝑧−𝑥𝑦 , (51)
1 + 𝑥𝑦

which after transformation gives the expression:

𝑦+𝑧 𝛼 𝑥𝑧−𝑦 𝛼
𝜇𝛼 − 𝑥+1
𝜈 + 𝑥+1
𝜎
𝛼
𝜀 = , (52)
𝑧
equivalent to thesis of theorem. ⊔
⊓

The space-like 4D relative velocity of Einstein–Oziewicz can now be assigned to the

4D time-like relative velocity (Minkowski type):
( )𝜇 𝜀𝜇 + 𝜎 𝜇
𝛽 𝛼 (𝜎, 𝜈, 𝜇) ∶= 𝜀(𝜎, 𝜈, 𝜇) ⟨+| 𝜎 = √ =∶ (𝜇 ⊟𝜎𝑀 𝜈)𝛼 . (53)
1 + 𝜀◦𝜀∕𝑐 2

The velocity of 𝛽 𝛼 will be called the Einstein-Minkowski relative velocity. Note that the
operation ⊟𝜎𝑀 which we will call (ternary or pseudo-binary) Minkowski velocities sub-
traction is an internal operation. Explicitly calculating the Einstein–Minkowski velocity
is not so difficult now, but it is related to the simplest form of the covariant Lorentz
transformation – therefore it will be given in the form of the following theorem:

Theorem 5 (Covariant passive Lorentz transformation of velocity). The time-like

Einstein–Minkowski relative velocity 𝛽 𝛼 is a passive Lorentz transformation (boost) of
the Minkowski velocity 𝜇𝛼 to a system with the Minkowski velocity 𝜈 𝛼 from the reference
system with the Minkowski velocity of 𝜎 𝛼 , and has the following form:

𝜇◦(𝜈 + 𝜎) 𝛼 𝜇◦𝜈 𝛼
𝛽 𝛼 (𝜎, 𝜈, 𝜇) = (𝜇 ⊟𝜎𝑀 𝜈)𝛼 = 𝐿𝑝𝛼𝛽 (𝜎, 𝜈)𝜇 𝛽 = 𝜇𝛼 − (𝜈 + 𝜎 𝛼 ) + 2 𝜎 . (54)
𝜎◦(𝜈 + 𝜎) 𝑐2
18 Grzegorz M. Koczan, GOL 2021 conference in memoriam Z. Oziewicz, Mexico

Proof. In order to apply the definition (53) we first calculate the square of the Einstein–
Oziewicz velocity norm:
𝑐6
𝜀◦𝜀 = − 𝑐2, (55)
(𝜇◦𝜈)2
which is the same as for the Oziewicz–Świerk–Bolós binary velocity and with the ac-
curacy to the sign as for the Einstein velocities composition (see Statement 2 in [17]).
Since 𝜇◦𝜈 > 0 then applying (53) to (47) we directly obtain the last formula (54) of the
thesis according to [23].
It remains to be explained that the Einstein–Minkowski velocity is a Lorentz boost
of velocity 𝜇𝛼 into the system with the velocity 𝜈 𝛼 (passive transformation 𝐿𝑝). Indeed
it is so, because Einstein–Minkowski velocity is a time-like 4D notation analogous to
velocity subtraction (45) or (46). Nevertheless, it is still worth making sure that we are
dealing with a passive (and not active) transformation, for which 𝜇𝛼 = 𝜈 𝛼 does not
change:
𝛽 𝛼 (𝜎, 𝜈, 𝜈) = 𝐿𝑝𝛼𝛽 (𝜎, 𝜈) 𝜈 𝛽 = 𝜈 𝛼 (56)
The condition of identity transformation also occurs when the passive Lorentz boost
does not change the main (selected) frame of reference 𝜈 𝛼 = 𝜎 𝛼 :

𝛽 𝛼 (𝜎, 𝜎, 𝜇) = 𝐿𝑝𝛼𝛽 (𝜎, 𝜎) 𝜇𝛽 = 𝜇𝛼 . (57)

On the other hand, a passive boost of nominally zero velocity 𝜎 𝛼 with velocity 𝜈 𝛼 gives
the opposite velocity in a three-dimensional sense 𝜈̃ 𝛼 = (𝛾𝑣 𝑐, −𝛾𝑣 𝑣):
⃗
𝜎◦𝜈 𝛼
𝛽 𝛼 (𝜎, 𝜈, 𝜎) = 𝐿𝑝𝛼𝛽 (𝜎, 𝜈) 𝜎 𝛽 = 2 𝜎 − 𝜈 𝛼 = 𝜈̃ 𝛼 , (58)
𝑐2
when active transformation would give ordinary velocity 𝜈 𝛼 . This fact concludes the
proof of the theorem. ⊓
⊔
On the basis of the covariant passive Lorentz transformation, one can easily structure
the inverse transformation and thus the active transformation. Despite the simplicity of
calculations, it is worth presenting these properties in the form of a theorem.
Theorem 6 (Covariant active and inverse Lorentz transformations). If the Lorentz
boost is parameterized with two Minkowski velocities 𝜎 𝛼 and 𝜈 𝛼 , then swapping these
velocities leads to an inverse transformation or equivalently converts the passive trans-
formation 𝐿𝑝 into the active transformation 𝐿𝑎 (and vice versa):

𝐿𝑝−1 (𝜎, 𝜈) = 𝐿𝑝(𝜈, 𝜎) = 𝐿𝑎(𝜎, 𝜈), (59)

𝐿𝑎−1 (𝜎, 𝜈) = 𝐿𝑎(𝜈, 𝜎) = 𝐿𝑝(𝜎, 𝜈). (60)

The active covariant Lorentz transformation of Minkowski velocity explicitly takes the
following form:
𝜇◦(𝜈 + 𝜎) 𝛼 𝜇◦𝜎 𝛼
𝐿𝑎𝛼𝛽 (𝜎, 𝜈) 𝜇𝛽 = 𝜇 𝛼 − (𝜈 + 𝜎 𝛼 ) + 2 𝜈 , (61)
𝜎◦(𝜈 + 𝜎) 𝑐2
which differs from the passive transformation only in the last expression.
 Relative binary and ternary 4D velocities in SR: I. Covariant Lorentz transformation 19

Proof. The inverse Lorentz transformation is obtained by using a boost with the velocity
𝜈̃ 𝛼 opposite in the three-dimensional sense to 𝜈 𝛼 :

𝜇◦(𝜈̃ + 𝜎) 𝛼 𝜇◦𝜈̃ 𝛼
(𝐿𝑝−1 )𝛼𝛽 (𝜎, 𝜈) 𝜇𝛽 = 𝐿𝑝𝛼𝛽 (𝜎, 𝜈)
̃ 𝜇𝛽 = 𝜇𝛼 − (𝜈̃ + 𝜎 𝛼 ) + 2 𝜎 . (62)
𝜎◦(𝜈̃ + 𝜎) 𝑐2
Using (58) and symbols from the proof of theorem 3, we write further the above expres-
sion:
2𝑥𝑦 − 𝑧 + 𝑦
(𝐿𝑝−1 )𝛼𝛽 𝜇𝛽 = 𝜇𝛼 − (2𝑥𝜎 𝛼 − 𝜈 𝛼 + 𝜎 𝛼 ) + 2(2𝑥𝑦 − 𝑧)𝜎 𝛼 , (63)
2𝑥 − 𝑥 + 1
which after transformations gives the formula:

𝑧 + 𝑦 − 2(𝑥 + 1)𝑦 𝛼 𝑧 + 𝑦 𝛼
(𝐿𝑝−1 )𝛼𝛽 𝜇𝛽 = 𝜇𝛼 − 𝜈 − 𝜎 , (64)
𝑥+1 𝑥+1
which is equal to the right side of the thesis (61) of the theorem. The inverse of the
Lorentz transformation automatically is the conversion of the passive transformation to
the active transformation (and vice versa). Therefore, the third thesis (61) is proved.
Now it is enough to note that the proven expression (61) differs from (54) only by
swapping 𝜎 𝛼 and 𝜈 𝛼 . This proves (59), (60) and ends the proof of the theorem. ⊔
⊓

The first conclusion from the above theorem is the possibility of determining the
internal addition in the Minkowski velocities set:
𝜇◦(𝜈 + 𝜎) 𝛼 𝜇◦𝜎 𝛼
(𝜇 ⊞𝜎𝑀 𝜈)𝛼 = 𝐿𝑎𝛼𝛽 (𝜎, 𝜈) 𝜇𝛽 = 𝜇𝛼 − (𝜈 + 𝜎 𝛼 ) + 2 𝜈 . (65)
𝜎◦(𝜈 + 𝜎) 𝑐2
The Minkowski velocities subtraction described earlier is the Einstein–Minkowski rela-
tive velocity, and the addition appears to be something else, but is not separately named
here and denoted by a new symbol. Surprisingly, however, it can be interpreted as a
relative velocity, but relative to another body in a different frame of reference:

(𝜇 ⊞𝜎𝑀 𝜈)𝛼 = (𝜇 ⊟𝜈𝑀 𝜎)𝛼 = 𝛽 𝛼 (𝜈, 𝜎, 𝜇). (66)

The above relation shows the flexibility and importance of the concept of Einstein–
Minkowski relative velocity in the context of Ternary Special Relativity.
An analogous second conclusion of the theorem will be the definition of addition as
opposed to subtraction that creates a space-like Einstein–Oziewicz velocity (expressed
in terms of Minkowski velocities):
( )𝛼
(𝜇[−]𝜎0 𝜈)
̃ 𝛼 =∶ (𝜇[+]𝜎0 𝜈)𝛼 ∶= 𝛽(𝜈, 𝜎, 𝜇) ⩤ 𝜎 . (67)

Both ways of defining (on the left and on the right) are equivalent and lead to the formula:
[ 𝜇◦𝜎 𝛼 ] 𝛼
𝑐4 𝜇◦(𝜈 + 𝜎) 𝛼 𝛼
(𝜇[+]𝜎0 𝜈)𝛼 = 𝜇𝛼 − (𝜈 +𝜎 )+2 𝜈 −𝜎 . (68)
2(𝜈◦𝜎)(𝜇◦𝜎) − (𝜇◦𝜈)𝑐 2 𝜎◦(𝜈 + 𝜎) 𝑐2
The above formula, given without proof, no longer shows the simplicity and symmetry
of the previous formulas. However, the Einstein–Oziewicz velocity expressed by the
20 Grzegorz M. Koczan, GOL 2021 conference in memoriam Z. Oziewicz, Mexico

operation of Oziewicz–Einstein on binary velocities has a trivial difference between

addition and subtraction:

(𝜔01 ⊞𝜎0 𝜔02 )𝛼 ≡ (𝜔01 ⊞𝜎𝑂 𝜔02 )𝛼 ∶= (𝜔01 ⊟𝜎0 −𝜔02 )𝛼 ≡ (𝜔01 ⊟𝜎𝑂 −𝜔02 )𝛼 . (69)

At the end of Part I, the inverse velocity theorem for the Lorentz boost, which is the
key Ternary Special Relativity theorem, will be formulated and proven. It is not about
the opposite Minkowski velocity 𝜈̃ 𝛼 generating the inverse Lorentz transformation, but
the inverse velocity in terms of a parameter for a given Lorentz boost. The term inverse
velocity is borrowed from Oziewicz, who used it both in the context of binary relative
velocity and in the context of the Lorentz transformation leading to ternary velocity.
Theorem 7 (Inverse velocity of the covariant Lorentz transformation). If a passive
Lorentz boost 𝐿𝑝(𝜎, 𝜈) converts Minkowski velocity 𝜇 𝛼 to velocity 𝛽 𝛼 , then the parameter
of this boost, being Minkowski velocity, is expressed as follows :
( 2 )
(𝜇◦𝜎 + 𝛽◦𝜎)(𝜇𝛼 − 𝛽 𝛼 ) + 2 (𝛽◦𝜎)
𝑐2
+ 𝜇◦𝛽 − 𝑐 2 𝜎 𝛼
𝐿𝑝𝛼𝛽 (𝜎, 𝜈)𝜇𝛽 = 𝛽 𝛼 → 𝜈 𝛼 (𝜎, 𝛽, 𝜇) = ,
2(𝜇◦𝜎)(𝛽◦𝜎)∕𝑐 2 − 𝜇◦𝛽 + 𝑐 2
(70)
and can be treated as some kind of subtraction of 𝜇𝛼 and 𝛽 𝛼 , just as 𝛽 𝛼 is a specific
subtraction of 𝜇𝛼 and 𝜈 𝛼 .
Proof. We need to solve the equation (54) for 𝛽 𝛼 treating 𝜈 𝛼 as the unknown. Starting
from the general ternary formula (2):

𝜈 𝛼 (𝜎, 𝛽, 𝛼) = 𝑓2 𝜇𝛼 − 𝑓1 𝛽 𝛼 − 𝑓0 𝜎 𝛼 , (71)

it is worth noting that 𝑓2 = 𝑓1 (condition viii. holds). The justification for this fact is
⃗ 𝑢] = 𝑏⃗ then 𝐿𝑝(−𝑣)[
the simplest in 3D. It is known that if 𝐿𝑝(𝑣)[⃗ ⃗ = 𝑢⃗. This means
⃗ 𝑏]
the antisymmetry of the three-dimensional inverse velocity 𝑣( ⃗ 𝑢⃗) = −𝑣(⃗
⃗ 𝑏, ⃗ which is
⃗ 𝑢, 𝑏),
equivalent to 𝑓2 = 𝑓1 .
So it is enough to write the equations for the two unknown coefficients 𝑓0 and 𝑓1 .
For this, we calculate the two scalar products resulting from the formula 𝛽 𝛼 (54):

𝛽◦𝜎 = 𝜇◦𝜈, (72)

𝛽◦𝜈 = 2(𝜇◦𝜈)(𝜈◦𝜎)∕𝑐 2 − 𝜇◦𝜎. (73)

By inserting the formula (71) into these two equations and using simplifying symbols
𝑏 ∶= 𝛽◦𝜎∕𝑐 2 , 𝑚 ∶= 𝜇◦𝜎∕𝑐 2 , 𝑛 ∶= 𝜇◦𝛽∕𝑐 2 we get:

𝑏 = (1 − 𝑛)𝑓1 − 𝑚 𝑓0 , (74)
( )
(𝑛 − 1)𝑓1 − 𝑏 𝑓0 = 2𝑏 (𝑚 − 𝑏)𝑓1 − 𝑓0 − 𝑚. (75)
The solution of this linear system of equations (e.g. by the determinant method) leads,
after simplification, to the result:
𝑏+𝑚
𝑓1 = , (76)
1 − 𝑛 + 2𝑏𝑚
 Relative binary and ternary 4D velocities in SR: I. Covariant Lorentz transformation 21

1 − 𝑛 − 2𝑏2
𝑓0 = , (77)
1 − 𝑛 + 2𝑏𝑚
which is equivalent to the thesis of the theorem. ⊔
⊓

In the case of the inverse velocity for the active transformation 𝐿𝑎(𝜎, 𝜈) it is enough
to replace the arguments in the resulting formula 𝜈 𝛼 (𝜎, 𝛽, 𝜇) ⟶ 𝜈 𝛼 (𝜎, 𝜇, 𝛽).
The Oziewicz–Ungar–Dragan ternary relative velocity from the previous section is
based on the above theorem. The application of the covariant Lorentz transformation
allowed for a significant simplification of the proof of the theorem. For example, in
Dragan [10], the calculations extend through three subsections (it does not apply to the
"magic four-rule"), and the previous relatively simple heuristic derivation of the author
[17] was not fully general and had to be based on a computationally complex proof of
the relevant lemma. However, the 3D derivation presented in the proof of Theorem 3
turned out to be the simplest.

Conclusions of Part I. The only covariant relative velocity that does not depend on the
selected frame of reference is Oziewicz–Świerk–Bolós (and Matolcsi or Bini–Carini–
Jantzen) velocity. In the sense of space-like velocity, it is a standard velocity, because
it is normalized to the real speed of relative motion (Minkowski time-like 4D velocity
is normalized to the speed of light). Nevertheless, the composition of binary velocities
has a ternary character, i.e. it additionally depends on the selected reference system. It
turns out that virtually every comprehensive relation of motion is ternary, starting with
the Lorentz transformation of velocity. The truly covariant Lorentz transformation leads
to a space-like relative velocity of Einstein–Oziewicz, which is ternary in nature (or at
least pseudo-binary). Also the time-like equivalent of this velocity (called the Einstein–
Minkowski velocity) is characterized by a certain ternarity.
However, the role of the canonical ternary velocity is played by the relative velocity
of Oziewicz–Ungar–Dragan (also Celakoska–Chakmakov–Petrushevski and Urbantke),
which is based on the so-called inverse velocity in terms of the covariant Lorentz trans-
formation parameter. The relative ternary velocity of Oziewicz–Ungar–Dragan is the
only one of the considered ones to be characterized by antisymmetry, which seems oblig-
atory for relative velocities. Composition of ternary velocities in the Oziewicz sense
takes the form of Einstein velocity composition generalized to 4D. Surprisingly, such
folding is not ternary – it does not depend on the third parameter, other than on the two
folding velocities. However, this Oziewicz–Einstein ternary composition is not a fully
general case of composition of relative ternary velocities based on transitivity. This fact
means that Oziewicz’s research program, called here Ternary Special Relativity (TSR),
is rich enough and still open.
Another open issue for TSR remains the form of associativity composition of veloc-
ities. As an example of such composition, the clarified composition of binary velocities
in the Oziewicz sense, based on transitivity, was given.

About Part II. The development of the foundations of Oziewicz’s research program
(Ternary Special Relativity – TRS) turned out to be so rich that the results of a slightly
22 Grzegorz M. Koczan, GOL 2021 conference in memoriam Z. Oziewicz, Mexico

different nature had to be moved to Part II. Nevertheless, the results of Part II will con-
cern the formalism defined in Part I.
Part II of the work will be introduces two 4D ternary generalizations of relative bi-
nary velocity called pseudo-binary relative velocities: cross and axial. The first pseudo-
binary cross velocity is a proper 4D generalization of the author’s 3D jet velocity (3D bi-
nary velocity). The second is a 4D generalization of the 3D axial velocity of Fernándeuz-
Guasti and the author. The cross velocity is a relatively simple modification of the binary
velocity, while the second pseudo-binary axial velocity is a bit more complicated – al-
most like typical ternary velocity. Despite this complexity, axial velocity is a straightfor-
ward direct generalization of Einstein composition of velocities in one spatial dimension
into the general case of 4D space-time. Additionally, the problem of general composi-
tion of ternary velocities will be attacked – more general than composition in the sense
of Oziewicz–Einstein.
Probably the Part II of the work will be presented at the next GOL conference in
memory of Zbigniew Oziewicz, and the announced results have been written and calcu-
lated in handwritten notes.

References

1. Abramowicz, M.A.: Velocity, acceleration and gravity in Einstein’s relativity.

arXiv:1608.07136 (2016).
2. Bażański, S.L.: Decomposition of the Lorentz Transformation Matrix into Skew-Symmetric
Tensors. Journal of Mathematical Physics 6:1201–1203 (1965). doi:10.1063/1.1704760
3. Bini, D., Carini, P., Jantzen, R.T.: Relative observer kinematics in general relativity. Classical
and Quantum Gravity 12:2549–2563 (1995). doi:10.1088/0264--9381/12/10/013
4. Bolós, V.J.: Intrinsic definitions of relative velocity in general relativity. Comm Math Phys
273:217–36 (2007), arXiv:0506032v1 (2005). doi:10.1007/s00220-007-0248-9
5. Celakoska, E.G.: On isometry links between 4-vectors of velocity. Novi Sad Journal of Math-
ematics 38(3):165–172 (2008).
6. Celakoska, E.G., Chakmakov, D.: Lorentz Link Problem and Solutions. Proceedings of
the Fourth International Scientific Conference "Faculty of Mathematics and Natural Sci-
ence", South-West University "Neofit Rilski", Blagoevgrad, Bulgaria, FMNS2011 Vol.1:16–21
(2011).
7. Celakoska, E.G., Chakmakov, D.: Mathematical model of relativistic 3-acceleration. Interna-
tional Journal of Advances in Applied Mathematics and Mechanics 6(2):14–19, ISSN: 2347-
2529 (2018).
8. Celakoska, E.G., Chakmakov, D., Petrushevski, M.: On Parameterization of Lorentz Boost
Links. International Journal of Contemporary Mathematical Sciences 10(2):85–90 (2015). doi:
10.12988/ijcms.2015.513
9. Choi, Y.-H.: Uniqueness of the isotropic frame and usefulness of the Lorentz transformation.
Journal of the Korean Physical Society 72(10):1110–1120 (2018). doi:10.3938/jkps.72.1110
10. Dragan, A.: Niezwykle Szczególna Teoria Względności. Roz. 3. Obrót Thomasa–
Wignera. Roz. 8.2. Magiczna czteroreguła (Unusually Special Theory of Rel-
ativity. Chap. 3. Thomas–Wigner Rotation. Chap. 8.2. The magic four-rule).
Monograph – lecture notes: www.researchgate.net/publication/ 265887295 (2012),
www.fuw.edu.pl/dragan/Fizyka/Nstw.pdf (2013).
11. Dragan, A.: Unusually Special Relativity. World Scientific (2021). doi:10.1142/q0319
 Relative binary and ternary 4D velocities in SR: I. Covariant Lorentz transformation 23

12. Einstein, A.: Zur elektrodynamik bewegter Körper (On the Electrodynamics of Moving Bod-
ies). Annalen der Physik 17:891–921 (1905).
13. Fahnline, D.E.: Manifestly covariant, coordinate-free dyadic expression for planar homo-
geneous Lorentz transformations. Journal of Mathematical Physics 24:1080–1086 (1983).
doi:10.1063/1.525833
14. Hestenes, D.: Proper particle mechanics. Journal of Mathematical Physics 15:1768–1777
(1974).
15. Grasso, M., Korzyński, M., Serbenta, J.: Geometric optics in general relativity using bilocal
operators. Physical Review D 99, 064038:1–32 (2019).
16. Kocik, J.: Making sense of relativistic composition of velocities. arXiv:1910.06785v1 (2019).
17. Koczan, G.M.: Relativistic Relative Velocity and Relativistic Acceleration. Acta Physica
Polonica A. 139(4):401–406 (2021). doi:10.12693/APhysPolA.139.401
18. Koczan, G.M.: New definitions of 3D acceleration and inertial mass not violating F=MA
in the Special Relativity. arXiv:1909.09084v1 (2019). Results in Physics 24(5), 104121:1–19
(2021). doi:10.1016/j.rinp.2021.104121
19. Koczan, G.M.: The new definition of three-dimensional relativistic acceleration and
its consequences within the SR. Poster for 46th Extraordinary Congress of Polish
Physicists on the 100 Years of the Polish Physical Society, Warsaw, Poland, Oc-
tober 16–18 (2020). https://100lat.ptf.net.pl/plakaty/0286-plakat_en- 1be04124.pdf,
www.researchgate.net/publication/344670531. doi:10.13140/RG.2.2.27955.45603
20. Koczan, G.M.: Linear geometrical constructions of the correspondence of 4D and
3D vectors in SR, containing covariant ternary relativistic acceleration. Poster for
47th Congress of Polish Physicists, Bydgoszcz, Poland, September 19–23 (2021).
www.researchgate.net/publication/354692005. doi:10.13140/RG.2.2.33945.95846.
21. Koczan, G.M.: Operator Położenia w Relatywistycznej Mechanice Kwantowej (Position Op-
erator in Relativistic Quantum Mechanics). Master’s thesis University of Warsaw (2002).
https://www.fuw.edu.pl/gkoczan/INNE/mgr.pdf.
22. Krause, J.: Lorentz transformations as space-time reflections. Journal of Mathematical
Physics 18:889–893 (1977). doi:10.1063/1.523355
23. Krause, J.: Lorentz transformations as space–time reflections. II. Timelike reflections. Journal
of Mathematical Physics 19:370–374 (1978). doi:10.1063/1.523680
24. Matolcsi, T.: Spacetime without reference frames. Akadémiai Kiadó, Budapest (1993), ISBN
963 05 6433 5.
25. Matolcsi, T., Goher, A.: Spacetime without reference frames: An application to the velocity
addition paradox. Studies in History and Philosophy of Modern Physics 32(1):83–99 (2001).
26. Oziewicz, Z.: How do you add relative velocities?. Group Theoretical Methods in Physics,
Conference Series 185:439–444, CRC Press (2004). doi:10.1201/9781482269185
27. Oziewicz, Z.: Relativity groupoid, instead of relativity group. International Jour-
nal of Geometric Methods in Modern Physics 04 (05):739–749 (2007). doi:10.1142/
S0219887807002260
28. Oziewicz, Z.: Ternary relative velocity. arXiv:1104.0682v1 (2011).
29. Oziewicz, Z.: Relativity without Lorentz group. Academia (2005/2011):
https://www.academia.edu/17230451/Special_relativity_without_Lorentz_group1.
30. Oziewicz, Z.: Æther Needs Relativity Group: Lorentz Group if and only if Æther. Academia
(2006/2008): https://www.academia.edu/12338869/Æther_Needs_Relativity_Group
_Lorentz_Group_if_and_only_if_Æther.
31. Oziewicz, Z., Page, W.S.: Concepts of relative velocity. arXiv:1104.0684v1 (2011).
32. Oziewicz, Z.: Are peer reviewers Guardians of the Truth?. Proceedings of the Natural Phi-
losophy Alliance, 20th Annual Conference of the NPA 10:228–231, College Park, Maryland,
USA (2013).
 24 Grzegorz M. Koczan, GOL 2021 conference in memoriam Z. Oziewicz, Mexico

33. Oziewicz, Z.: Science must never be objective. Proceedings of the XVI-th International Con-
ference, The Science and Quality of Life, Vilnius, Lithuania, Studium Vilnense A 10:3–8
(2013), ISSN 1648-7907.
34. Sbitneva L.: Differential equations of smooth loops related to some space-time mod-
els: Integrability conditions and geometry. Journal of Knot Theory and Its Ramifications
27(7):1841004 (2018), World Scientific.
35. Szostek, R.: Derivation of all linear transformations that meet the results of Michelson-
Morley’s experiment and discussion of the relativity basics. Moscow University Physics Bul-
letin 75(6):684–704 (2020), ISSN: 0027-1349. doi:10.3103/S0027134920060181
36. Świerk, D.A.: Teoria względności i struktury produktowe (Relativity theory and product
structures). Master Thesis supervised by Z. Oziewicz, University of Wrocław, Institute of The-
oretical Physics (1988), https://www.researchgate.net/publication/356144047.
37. Tsamparlis, M.: Special Relativity – An Introduction with 200 Problems and Solutions
(Chap. 15. The Covariant Lorentz Transformation). Springer (2010, 2019). doi:10.1007/
978-3-642-03837-2
38. Ungar, A.A.: The Relativistic Noncommutative Nonassociative Group of Velocities and the
Thomas Rotation. Results in Mathematics 16:168–179 (1989).
39. Ungar, A.A.: A note on the Lorentz transformations linking initial and final four-vectors.
Journal of Mathematical Physics 33(84) (1992).
40. Ungar, A.A.: Beyond the Einstein Addition Law and its Gyroscopic Thomas Precession: The
Theory of Gyrogroups and Gyrovector Spaces. Kluwer Academic Publishers (2001,2002). doi:
10.1007/978-94-010-9122-0
41. Ungar, A.A.: Analytic hyperbolic geometry: Mathematical foundations and applications.
World Scientific Publishing (2005).
42. Ungar, A.A.: The relativistic proper-velocity transformation group. Progress In Electromag-
netics Research 60:85–94 (2006). doi:10.2528/PIER05121501
43. Ungar, A.A.: Gyrogroups, the grouplike loops in the service of hyperbolic geometry and
Einstein’s Special Theory of Relativity. Quasigroups and Related Systems 15:141-168 (2007).
44. Urbantke H.K.: Lorentz Transformations from Reflections: Some Applications. arXiv:math-
ph/0212038v1 (2002). Foundations of Physics Letters 16(2):111–117 (2003).
45. van Wyk, C.B.: Lorentz transformations in terms of initial and final vectors. Journal of Math-
ematical Physics 27(5):1306–1314 (1986). doi:10.1063/1.527136
46. van Wyk, C.B.: The Lorentz operator revisited. Journal of Mathematical Physics 32(2):425–
430 (1991). doi:10.1063/1.529429

View publication stats