IMPROVEMENTS OF SOME OF THE RELATIVITY EQUATIONS

Andrés L. Granados M., July, 2024.

UNIVERSIDAD SIMON BOLIVAR
Departamento de Mecánica

Several improvements of some relativity equations are presented in this article [Granados,2022]. The
main notation (symbolic) will be that used principally by Josiah Willard Gibbs (1839-1903). Where escalars
are small italic or greek letters (e.g. c,γ), vectors are bold roman or greek small letters (e.g. v,β) or bold
roman or calligraphic capital letters (e.g. F,F ), tensors are capital bold roman letters (e.g. T), and invariants
and other scalars and components are italic or calligraphic capital letters (e.g. R, K). Eventually, we use
Fraktur bold fonts (small & capital) for quaternion vectors (e.g. a,A).
We shall use dot ‘ . ’ for escalar multiplication of vectors (one index contraction), also for a multiplication
of a matrix (symbolized by a tensor of order two) by a vector (one index colateral contraction). Also we
shall use double dots ‘ : ’ for escalar multiplication of two tensors of order two (meaning also two index
colateral contractions). We distinguish other multiplications such as vector multiplication ‘ × ’, and tensor
multiplication ‘ ⊗ ’. This last applied to two vectors a, b we simplify as a dyadic a ⊗ b = ab.
Also we shall use matrix notation when it is refered to the components (in any coordinate system) of
a vector with { · } or a tensor of second order with [ · ].
The use of the aforementioned symbolic notation makes the found equations independent from the
coordinates system (including  and ⊥ to v system) and therefore more general.

Minkowski vs. Quaternion

Some typical quantities are mass m (at rest with subindex ‘ o ’), linear momentum p, velocity u, Energy
E, and force F

m = γu mo γu = (1 − βu2 )−1/2 βu = u/c βu2 = βu .βu (1)

dp
p = mu F= E = m c2 K = E − Eo (2)
dt
being Eo = mo c2 the energy at rest and K the kinetic energy. The relativistic fictitious mass m depends
on the velocity through γu , but this Lorentz factor should accompany to the velocity in Minkowski velocity
instead.
In relativity they are defined the Minkowski velocity (γu u = dr /dt = dr/dt is the velocity observed
by a moving observer at velocity u = dr/dt, due to shrinking of length and time dilatation according to the
Lorentz transformation), the Minkowski moments and the Minkowski forces as follows
     
iγu c iE/c iγu Ė/c d℘
{υ} = {℘} = {mo υ} = {F } = F= (3)
γu u p γu F dτ

being the derivative calculated with respect to the proper time τ ( dt = γu dτ ). The quantities of Minkowski
satify the following relations

E2
υ.υ = −c2 F .υ = 0 ℘.℘ = −m2o c2 = p.p − (4)
c2
√
because the time components in those vectors are imaginary (multiplied by i = −1, the imaginary base)
On the other hand the quaterniones are tetra-dimensional vectorial quantities due to William Rowan
Hamilton (1805-1865) where any vector has the components in the base IB = {1, i, j, k} (one real in time

1
and three imaginary in space, a∗ is the dual of a, a tetra-dimensional generalization of the bi-dimensional
complex numbers and their conjugated)

a = a0 + a = a0 + a1 i + a2 j + a3 k a∗ = a0 − a = a0 − a1 i − a2 j − a3 k (5)

with the next properties (a0 and a0 is the same saclar quantity)
⎧
⎪
⎪ i ∗ j = −j ∗ i = k
⎨
i ∗ i = j ∗ j = k ∗ k = i ∗ j ∗ k = −1 j ∗ k = −k ∗ j = i (6)
⎪
⎪
⎩
k ∗ i = −i ∗ k = j

(add 1 ∗ a = a ∗ 1 = a, ∀a ∈ H) where the product ‘ ∗ ’ is an special non-commutative multiplication in

the Clifford algebra over the field of H = C2 , set of quaternions [Hamilton,1899]. Similarly as the vectorial
multiplication ‘ × ’ in the three-dimensional space (see (11) below).
This permits us to define the components contravariant and covariant of a quaternion A

{A}IB = {Aα } = {A0 , A} {A∗ }IB = {Aα } = {A0 , −A} Aα = gαβ Aβ (7)

where A corresponds to the components purely spatial (imaginary three-dimensional A = A1 i + A2 j + A3 k.

gi,j = 0 for i = j, g00 = 1, gii = −1 for i = 0). Under any components transformation

∂xα β ∂xβ
Aα = A Aα = Aβ (8)
∂xβ ∂xα

the scalar multiplicaction of two treta-vectors quaternions is invariant

∂xβ ∂xα ∂xβ

A∗ .B = A .B  = α γ
Aβ B γ = Aβ B γ = δγβ Aβ B γ = Aα B α = A.B (9)
∂x ∂x ∂xγ

therefore, according to the convention (5) and (7), we have (B ∗ B∗ = B.B = B2 )

A∗ .B = A.B∗ = A.B = Aα B α = Aα Bα = A0 B0 − A.B (10)

Finally, the multiplication ‘ ∗ ’ of two quaternions a and b is (in components)

     
a0 b0 a0 b0 − a.b
{a ∗ b} = ∗ = (11)
a b a0 b + b 0 a + a × b

Accordingly, the Minkowski quantities, velocity, linear momentum and force are defined in the following
form [Goldstein,1980]
     
γu c E/c γu Ė/c d℘
{υ} = {℘} = {mo υ} = {F } = F= (12)
γu u p γu F dτ

These quantities satisfy the following relations

E2
υ.υ = c2 F .υ = 0 ℘.℘ = m2o c2 = −p.p + (13)
c2

similar to (4).

2
 Particularly, in modern times, like Einstein in his time, the quaternions representation of space-time
quantities is prefered. In early times the index for time quantities was 4, the fouth dimension, but this has
been progressively changed to index 0, the primordial dimension (also the place into the vector, from the last
to the first).

Lorentz Transformation
La dilatation of time may be expressed from the line metric in space-time as [French,1968]

v2 1 v
(cdτ )2 = (c dt)2 − (v dt)2 = (c dt)2 1 − 2 dt = γ dτ γ= β= (14)
c 1 − β2 c

being τ the proper time that is the observed time following a moving particle at the velocity v respect to the
inertial system S of reference at rest.
The Lorentz transformation between two parallel systems of reference, the system S  moving at velocity
v in x direction respect to the system S at rest, is espressed by

ct = γ (ct − βx) ct = γ (ct + βx )

x = γ (x − βct) x = γ (x + βct )
(15)
y = y y = y
z = z z = z

wher exist an additional factor γ, defined by (14.c, d), and the time is not absolute as in the Galileo trans-
formation. The coincidence of the origins of both system has been chosen to simplify. The velocity v may be
so high as that for light c (never superior or equal, γ > 1), such that its relation β is not negligible. When
v c, then β ≈ 0 and γ ≈ 1, and in the limit there is obtained the Galileo transformation. In matrix
notation this transformation is expressed by
⎧ ⎫ ⎡ ⎤⎧ ⎫ ⎧ ⎫ ⎡ ⎤⎧ ⎫
⎪ ct γ −γβ 0 0 ⎪ ct ⎪ ct γ γβ 0 0 ⎪ ct ⎪
⎨ ⎪ ⎬ ⎨ ⎬ ⎪
⎨ ⎪ ⎬ ⎨ ⎬
x ⎢ −γβ γ 0 0⎥ x x ⎢ γβ γ 0 0 ⎥ x
 =⎣ ⎦ =⎣ ⎦  (16)
⎪
⎩ y ⎪⎭ 0 0 1 0 ⎪ ⎩y⎪⎭ ⎪
⎩y⎪ ⎭ 0 0 1 0 ⎪ ⎩ y ⎪⎭
z 0 0 0 1 z z 0 0 0 1 z

For any inclination of the velocity v, we have [Jackson,1999]


γ−1
ct = γ (ct − β.r) θ=
β2
γ−1 (17)
r = r + ββ. r − γβ ct
β2
= r + θ ββ. r − γβ ct

where

v v 1
ê = β= = β ê β = β = β2 γ= β 2 = β.β (18)
v c 1 − β2

For the tetra-dimensional space it is defined the tetra-vector

 
ct
{x} = (19)
r

3
and the Lorentz transformation, whose matrix id aymmetrical, it is expressed for orthonormal systems as
[Møller,1952;pp.40-41]
⎡ ⎤
γ −γβx −γβy −γβz  
 ⎢ −γβx 1 + θ βx2 θ β x βy θ β x βz ⎥ γ −γβ
x = L.x [L] = ⎣ ⎦= (20)
−γβy θ β y βx 1 + θ βy2 θ β y βz −γβ I + θββ
−γβz θ β z βx θ β z βy 1 + θ βz2

where θ = (γ − 1)/β 2 and the low-right sub-matrix in the corner may be replaced by I + (γ − 1) êê (I
is the identity matrix). The inverse transformation is obtained changing the sign of β by relativity, what
changes only the sign of the first row and the first column in their spatial part. A substitution of the Lorentz
transformation in the differential of the proper time gives again that they are equivalent in both systems and
it is an invariant of the transformation. In the case of the propagation of the light the differential of proper
time is zero (see(14)).

Velocity Sum
The application of two consecutive Lorentz transformations, one with velocity v1 and the other with
velocity v2 results in an equivalent transformation of velocity v, all of these velocities being collinear. The
result is a Lorentz transformation of velocity
v1 + v2 β1 + β 2
v= β= (21)
1 + v1 v2 /c2 1 + β1 β 2
equivalents between them. If either velocity v1 or v2 is equal to c, then the result is that v is equal to c.
According to this rule for adding speeds of Lorentz transformations, it is impossible to obtain speeds equal
to or greater than the speed of light (v < c) and β is always less than unity.
Sean ux = dx/dt, uy = dy/dt y uz = dz/dt las componentes de la velocidad u de una partı́cula obser-
vada desde un sistema inercial S en reposo. Sean ux = dx /dt, uy = dy  /dt y uz = dz  /dt las componentes
de la velocidad u de la misma partı́cula observada desde un sistema inercial S  paralelo, moviéndose con
velocidad v en x respecto a S. A partir de las transformación de Lorentz (15) se pueden obtener las relaciones
entre ambas velocidades como [French,1968]

ux − v ux + v
ux = ux =
1 + ux v/c2
1 − ux v/c2
uy uy
uy = uy = (22)
γ (1 − ux v/c2 ) γ (1 + ux v/c2 )
uz uz
uz =
γ (1 − ux v/c2 ) uz =
γ (1 + ux v/c2 )

where it is established that the composition of the velocities u and v = v ê gives a speed u that in in no case
should it exceed the value of c (in its norm).
When the speed u with respect to the system S  is known and the same speed with respect to the
system S is desired, knowing that S  moves with respect to S at a speed v, then the result is the following
u + v u⊥
u = u⊥ = (23)
1 + u .v/c2 γ (1 + u .v/c2 )

and in function of their respective γu , γu and γv = γ [Jackson,1999;Chp.11,p.532], so that there are no
doubts when applying (18),

γu u = γu γv (u + v) [ I − (γ − 1)/γ êê ] . u =


γu u⊥ = γu u⊥ u + v (24)
[ I − γ/(1 + γ) ββ ] . u =

γu = γu γv (1 + u .v/c ) 2 γ (1 + u .v/c2 )

4
The notation u = u.ê and u⊥ = u − u ê refer to the components from the parallel and perpendicular
velocity u, respectively, to the velocity v = v ê, where v = v. As can be seen, the expressions (23) − (24)
are the generalization of the expressions on the right side of (22). The last expression of (24) uses the identity
(32.b) in its deduction, and the novelty in this case is the innecessary separation of parallel and perpendicular
velocity components. The inverse composition of the velocities, that is, the velocity components u as a
function of the velocity components u, is obtained by changing the sign to v.

Aceleration Sum
Again in the case where S  moves with velocity v in x with respect to S, the accelerations are trans-
formed in a way special according to the following expressions [French,1974]
 
1  uy (ax v/c2 )
ay = ay −
ax [ γ (1 + ux v/c2 ) ]2 1 + ux v/c2
ax =  2 3   (25)
[ γ (1 + ux v/c ) ] 1  uz (ax v/c2 )
az = az −
[ γ (1 + ux v/c2 ) ]2 1 + ux v/c2
where the differentials of (22) or (23) have been calculated for the velocities and (15.e) for the time ( ct =
γ(ct + βx ) ), keeping v and v constant, with the objective of obtaining them (Note: dt = γ (1 + u .v/c2 ) dt ).
En el caso general, donde S  se mueve con velocidad v respecto a S, se distinguen una componente
paralela a = a.ê y otra perpendicular a⊥ = a − a ê (Note: a = du/dt, a = du /dt )
 
a 1  u⊥ (a .v/c2 )
a = a⊥ = a⊥ − (26)
[ γ (1 + u .v/c2 ) ]3 [ γ (1 + u .v/c2 ) ]2 1 + u .v/c2
he inverse transformations are obtained by changing the sign to v and v, considered constants, in the expres-
sions (11) and (12), respectively. Finally, the expressions (12) can be grouped into one, making a = a ê + a⊥
and a = a ê + a⊥ , with which you obtain

a + γ a⊥ [ (γ − 1)/γ + u .v/c2 ] − γ u⊥ (a .v/c2 )

a= (26 )
[ γ (1 + u .v/c2 ) ]3
Do not confuse the acceleration a with the inertial acceleration (for the mass mo ) γun a = γu [ I + γu2 βu βu ] . a
obtained ahead in (32). Again, the novelty in this case is the innecessary separation of parallel and perpen-
dicular acceleration components.

Complete New Deduction of E = mc2

The original Einstein [1905] deduction is reproduced almost exactly in the same way in [Goldstein,1980].
Most of the others deduction [Sokolnokoff,1951;p.272] assume the innecessary assumption that F is parallel to
a (and u, or βu  β̇u ), which in relativity mechanics is not certain at all as it should be seen in (32.a) where
the intercalated matrix is not the identity matrix as in the newtonian mechanics, but is a symmetrical matrix
that means there is a parallel and a perpendicular force components (analized in (31) below). The following
deduction resolve this incongruence intercalating an escalar multiplication between F and u in the initial
kinetical energy integral. This operator ‘ . ’ is dragged up to the final in all the terms (including βu .β̇u = β β̇)
of the following deduction
 r  t  t  t
d dγu du
K= F . dr = F . u dt = (γu mo u) . u dt = mo u . u + γu u . dt
ro to to dt to dt dt
 t  t
2 dγu 2
= mo c βu + γu βu . β̇u dt = mo c2 γu ( γu2 βu2 + 1 ) βu . β̇u dt
to dt to (27)
 t  βu  γu
= m o c2 γu3 βu . β̇u dt = mo c2 γu3 βu . dβu = mo c2 dγu = mo c2 (γu − 1)
to βo γo
2 2
= mc − mo c = E − Eo Ė = K̇ = F.u E 2 = p2 c2 + m2o c4

5
where is evident the use of the identities
γ̇u = γu3 βu . β̇u 1 + γu2 βu2 = γu2 (28)
at he initial developement (for γ̇) and at the final regression (third last term).

Force and Acceleration

According to the equations (1) and (2), the force F is calculated by (use (28.a))
d du dγu du du
F= (γu mo u) = γu mo + mo u = γu mo + γu3 mo βu βu . (29)
dt dt dt dt dt
In other words, it may be said, using the acceleration a = du/dt and (2), that (use (28.b))

F = γu2 m a (when a  βu ) (30)

Some authors [French,1968] [Jammer,1957] [Ganley,(1963)] justify that the expression (30) is only valid for the
parallel component a of the motion when βu  a , as indicated by the last two factors of the last term. But
for the transverse component a⊥ , the second term of the second and third members of (29) must not intervene
(see the scalar product operation in the last member of (29)), because they cancel out when βu ⊥ a⊥ , which
is equivalent to saying that the mass remains unchanged (dγu /dt = 0) in a transverse impulse. Therefore,
under this argument, we have

F = F. êu = γu2 m a F⊥ = F − F êu = [ I − êu êu ] . F = m a⊥ êu = u/u (31)

or equivalently ( a = a.êu , a⊥ = a−a êu )

F = F êu + F⊥ = m [ I + (γu2 − 1) êu êu ] . a = m [ I + γu2 βu βu ] . a (γu2 − 1) êu êu = γu2 βu βu (32)
and (global) force and acceleration do not have the same directions.“In contrast to the Newtonian conception,
it is easy to demonstrate that in relativity the magnitude of force does not have the same direction, in
general, as the acceleration it produces...” [Jammer,1957]. Which contradicts (30) in general cases, being
valid only when the parenthesis condition is satisfied (compare (29) with (32.a) ). Do not confuse this
inertial acceleration (for mass mo ) [γun ].a = γu [ I + γu2 βu βu ] . a, with the acceleration a at (26 ) seen from
the reference system S  . The identity (32.b) is a generalization of the identity (28.b) (valid only when êu is
the base vector of the principal axis).
Due to the invariance that the expression (2.b) must maintain, through the Lorentz transformation L,
below is what the same expression looks like in the transformed variables (three-dimensional Ľ = I + θββ
contains only the spatial part of the elements of tetra-dimensional L)

dp dp Ľ.ṗ − γβ Ė/c [Ľ − γββu ].F

F= F = 
= 2
= (33)
dt dt γ (1 − u.v/c ) γ (1 − u.v/c2 )

where p and F, and likewise p and F , have been expanded in the first component of energy to be tetra-vectors
as in (3), (12) and (20), taking into account that d℘ = L.d℘ from the differential of

d℘ d℘
℘ = L.℘ F = = L. = L.F γu = γ γu (1 − u.v/c2 ) (34)
dτ dτ
(with c, v, β, γ and L being constants) and that dt = γ (1 − u.v/c2 ) dt from (15.a). Equivalently, dividing
the differential of the transformation of the momentums px , pyz in

E  /c = γ (E/c − β px ) E/c = γ (E  /c + β px )

px = γ (px − β E/c) px = γ (px + β E  /c)
(35)
py  = py py = py 
pz  = pz pz = pz 

6
by the differential of the time transformation in (15.a) already calculated above, we obtain

F − β (F.βu ) F⊥
F = F⊥ = (36)
1 − u.v/c2 γ (1 − u.v/c2 )

where (27.b), Ė/c = F.βu , has been substituted in the first expression. The following is also a satisfied
expression
F. (u − v)
F .u = (37)
1 − u.v/c2
which serves as the transformation equation for the power of energy (first component in the Minkowski force),
taking into account (27.b), such that F .u = Ė  , F.u = Ė and F.v = Fx v ( v aligned on x ).

Constant Force
Under the action of a constant force F on a particle of mass mo , it experiences an inertial acceleration
w (also constant) of the shape
⎡ ⎤
d ⎢⎢  u(t)
⎥
⎥= F =w d
⎣ (u γu ) = w = w êw (w = w) (38)
dt 2 ⎦ mo dt
u(t)
1− 2
c

Integrating this differential equation from certain initial conditions of position ro and velocity uo at t = 0,
we obtain

u γu − uo γuo = w t u γu = w t + zo = z(t) zo = z(0) = uo γuo uo = u(0) (39)

The auxiliary variable z(t) includes the effect of the acceleration wt plus the initial condition zo . This results
in the speed
 
z(t) 1 z2 z2o
u(t) =  γu =  = 1+ 2 γuo = 1+ (40)
z(t)2 u 2 c c2
1+ 2 1− 2
c c

The inclusion of this auxiliary variable z, which includes the initial condition for the velocity, allows γu to
now be expressed as a direct radical 1 + z2 /c2 , which was previously a disadvantage for the inverse radical
1/ 1 − u2 /c2 (the inner sign changes from ‘-’ to ‘+’ as well). For the position, the integration of u(t), the
following results [Dwight,1961;Int.200.01-201.01,pp.45-46] (combination of  and ⊥ of w parts from z)
 t   t
c2 z2 z z2 
c 
r(t) = ro + 1+ 2 êw + ln + 1+ 2  ( zo − (zo .êw )êw ) (41)
w c w c c 
o o

(z = x in integration formula in [ibid.], and 201.01 and 191.01 are equivalent, but z2 = z2 = x in the
last) for any instant t. In this result, the solution has been separated into two parts, the first, in the direction
of êw , constant, and the second, in the direction perpendicular to it, present only in zo for obvious reasons.
The proper time of the particle τ is calculated as [ibid.;Int.200.01,p.45]

    t
t t
z z2 
dt dt c 
τ = τo + = τo +  = τo + ln + 1+ 2  (42)
γu 2 w c c 
1 + z2
o o
o
c

7
where in the integration, the integration of the second previous part appears coincidentally. The inertial
acceleration w is calculated the same as before in (32) as

 3/2  1/2
z2 z2
w= [γun ].a = γu3 a + γu a⊥ = 1 + 2 a + 1 + 2 a⊥ ( ê = z/z )
c c (43)
= γu [ I + (γu2 − 1) êê ] . a = 1+ z2 /c2 2 2
[ I + (z /c ) êê ] . a

where the composition of a parallel and a perpendicular part again gives the same transformation as before
for the acceleration a = du/dt, but now expressed as a function of z, whose direction ê is the same as that
of u according to (39.b), but not necessarily the direction of êw (unless zo also points in that direction or is
null). As noted before, the ê direction of u and z does not coincide with the êw direction of F and w, nor it
is with the direction of a (solved from the last matrix equation a = [γun ]−1.w).

Reference
[1] Dwight, H. B. Tables of Integral and Other Mathematical Data, 4th Edition. The Macmillan
Company (New York), 1961.
[2] Einstein, A. “Does the Inertia of a Body Depend on Its Energy Content?”, Annalen der Physik,
Vol.18, pp.639-641, (1905).
[3] French, A. P. Special Relativity. W. W. Norton & Company, Inc. (New York), 1968.
[4] Jammer, M. Concepts of Force: A Study in the Foundations of Dynamics. Harvard University
Press (Cambridge), 1957. Harper (New York), 1962. Dover Publications (New York), 1999.
[5] Ganley, W.P. Am. J. Phys., Vol.31, pp.510-516, (1963).
[6] Goldstein, H. Classical Mechanics, 2nd Edition. Addison-Wesley, 1980.
[7] Granados M., A. L. Mecánica y Termodinámica de Sistemas Materiales Continuos, Funda-
mentos, Aplicaciones y Fenómenos. Universidad Simón Bolı́var (ISBN 980-07-2428-1), Departamento
de Mecánica, 1999. Corregido Agosto, 2022.
[8] Hamilton, W. R. Elements of Quaternions, Vols.I & II, Second Edition. Longmans, Green & Co.
(London), 1899.
[9] Jackson, J. D. Classical Electrodynamics, Third Edition. John Wiley & Sons (New York), 1999.
[10] Møller, C. The Theory of Relativity, Second Edition. Clarendon Press (Oxford), 1972.
[11] Sokolnikoff, I. S. Tensor Analysis-Theory and Applications. John Wiley & Sons, 1951.