In "CLASSICAL ELECTRODYNAMICS" by J.D.Jackson, 3rd Edition , $\S$ 11.3, the author gives in equation (11.19) a generalization of Lorentz transformation as follows :

If the axes in K and K' remain parallel, but the velocity $\:\mathbf{v}\:$ of the frame K' in frame K is in an arbitrary direction, the generalization of (11.16) is

$$ \begin{align} x'_{0} & =\gamma\left(x_{0}-\boldsymbol{\beta}\boldsymbol{\cdot}\mathbf{x}\right)\\ \mathbf{x}^{\prime} & = \mathbf{x} +\dfrac{\left(\gamma-1\right)}{\beta^{2}}\left(\boldsymbol{\beta}\boldsymbol{\cdot}\mathbf{x} \right)\boldsymbol{\beta}-\gamma\boldsymbol{\beta}x_{0} \end{align} \Biggr\} \tag{11.19} $$ where $$ \begin{align} \boldsymbol{\beta} & = \dfrac{\mathbf{v}}{c}\; \qquad \beta=|\boldsymbol{\beta}| \\ \gamma &=\left(1-\beta^2 \right)^{-1/2} \end{align} \tag{11.17} $$

and $$ \begin{align} x'_{0} & =\gamma\left(x_{0}-\beta x_{1}\right)\\ x'_{1} & =\gamma\left(x_{1}-\beta x_{0}\right)\\ x'_{2} & =x_{2}\\ x'_{3} & =x_{3} \end{align} \Biggr\} \tag{11.16} $$ the Lorentz Transformation with the velocity $\:\mathbf{v}\:$ parallel to the common $\:x-x'\:$ axis.

In case (11-16) it's permissible to talk about parallel axes. But in the generalized case (11-19) is it permissible to talk about parallel axes ? What is the meaning of parallelism in this later case ?

  • $\begingroup$ Relevant paper: aapt.scitation.org/doi/abs/10.1119/1.11632?journalCode=ajp $\endgroup$
    – PolyPhys
    Commented Oct 26, 2021 at 17:45
  • $\begingroup$ I once said that Lorentz boosts are interpreted as the coordinate transformations between observers with the same axes and you corrected me and said that there is no sense in "...between observers with the same axes..." generally. Would you be willing to elaborate if I post a question regarding this? $\endgroup$
    – Filippo
    Commented Aug 19, 2022 at 19:15
  • $\begingroup$ @Filippo : Take a look in my answer here How to represent a pair of inertial frames in relativity?. $\endgroup$
    – Frobenius
    Commented Aug 19, 2022 at 19:25
  • $\begingroup$ @Frobenius Thank you! $\endgroup$
    – Filippo
    Commented Aug 19, 2022 at 19:38
  • $\begingroup$ @Filippo : Welcome !!! $\endgroup$
    – Frobenius
    Commented Aug 19, 2022 at 19:50

3 Answers 3


In general, Lorentz (or rather Poincare) transformation are those transformations that keep the speed of light in any reference frame the same. They can be decomposed into the following:

1.) Translations
2.) Rotations and
3.) Boosts

Translations and Rotations are defined just like in the Galileo case and don't make up the "interesting" physics of special relativity. Therefore, one often only really talks about the boosts when talking about Lorentz transformations. These boosts can be done in some coordinate direction, most easily in the direction of an coordinate axis, for example the $x$-axes. The corresponding transformation is written in your equation (11.16). Now Jackson talks about a generalization, by that he means here to write down the transformation for a boost in $\overrightarrow{v}$-direction which he does in (11.19).
In principle it's the same kind of transformation. So in the same fashion as for the boost in $x_1$ direction, the spatial coordinate axes are parallel, i.e. the transformation does not include any rotation. If you were to shoot an arrow in $\overrightarrow{x}_i$-direction, $i=1,2,3$, the observer in the reference frame with primed coordinates would also observe the arrow going in $\overrightarrow{x}'_i$-direction.
The parallelity refers to the spatial coordinates only.

  • $\begingroup$ well, try to solve it yourself: suppose you have a set of events given in coordinates of $K$, maybe as a finite number of lights, say ten, blinking at time $t=0$ placed along the $x_0$ axis at a distance of 1, $\{(0,n,0,0) : n=1,2,...,10\}$. What do the events look like in reference frame $K'$? Will the primed observer see the lights flashing at the same time? $\endgroup$ Commented May 8, 2016 at 20:56
  • $\begingroup$ For parallelity, recall the example of the arrow I have given. I think your problem lies with the idea of "simultanous". Given the set I specified before, the events representing a array of lights flashing at $t=0$. As you said, they will not flash simultanously in the primed reference frame, they will however still be aligned along the $x'_1$-axis. Even though the do not blink simultanously in the primed reference system, you can think of a scenario where they turn on at the specified events. So they will appear, one after one another, and stay on, spatially parallel to the coordinate axis. $\endgroup$ Commented May 9, 2016 at 12:20
  • $\begingroup$ If you want to further discuss this, I'd prefer a more respectful writing. Being downvoted and writtenly yelled at, I feel more ad more discourage to spend time thinking about this with you and trying to formulate my thoughts. As what my example concerns, yes you are right, I only thought about boosts in $x$ direction. As what the parallelity regards, I think what is ment is: $\endgroup$ Commented May 9, 2016 at 14:42
  • $\begingroup$ Both (spatial) reference frames are embedded in $\mathbb{R}^3$ with unit vectors given by $\overrightarrow{e}_i$, e.g. $\overrightarrow{e}_1=(1,0,0)$. The same holds vor the primed $\overrightarrow{e}'_i $. In that sense, the axes are parallel, even though from the point of the observer, the other reference frame's axes might not seem parallel to the own ones. Might that be it? $\endgroup$ Commented May 9, 2016 at 14:42

See: wikipedia:....

$\mathbf{r}=\mathbf {r} _{\perp }+ \mathbf {r} _{\parallel }\;\;, \;\mathbf{r'}=\mathbf {r'} _{\perp }+ \mathbf {r'} _{\parallel }$

then the transformations are:$$\begin{cases} t'=\gamma(t-\frac{\mathbf {r}_{\parallel }.\mathbf{v} }{c^{2}} )\\\mathbf {r'}_{\parallel }=\gamma(\mathbf {r}_{\parallel }-\mathbf{v}t)\\\mathbf{r'}_{\perp } =\mathbf {r} _{\perp } \end{cases}$$


Introducing a unit vector $\mathbf{n}=\mathbf{v}/v=\boldsymbol{\beta}/ \beta \;$ in the direction of relative motion, the relative velocity is $\mathbf{v} = v\mathbf{n}$ with magnitude $v$ and direction $\mathbf{n}$, and vector projection and rejection give respectively $${\displaystyle \mathbf {r} _{\parallel }=(\mathbf {r} \cdot \mathbf {n} )\mathbf {n} \,,\quad \mathbf {r} _{\perp }=\mathbf {r} -(\mathbf {r} \cdot \mathbf {n} )\mathbf {n} }\;\;\;\;$$(*)

Accumulating the results gives the full transformations,

$$\begin{align} t' & =\gamma\left(t-\frac{v\boldsymbol{n}\boldsymbol{\cdot}\mathbf{r}}{c^{2}}\right)\\ \mathbf{r}^{\prime} & = \mathbf{r} +\left(\gamma-1\right)\left(\boldsymbol{r}\boldsymbol{\cdot}\mathbf{n} \right)\boldsymbol{n}-\gamma vt\boldsymbol{n} \;\;\end{align} \Biggr\} $$ with $x_{0}=ct\;,\;x'_{0}=ct'\;,\;\mathbf{r}\equiv \mathbf{x}$

$$\begin{align} x'_{0} & =\gamma\left(x_{0}-\boldsymbol{\beta}\boldsymbol{\cdot}\mathbf{x}\right)\\ \mathbf{x}^{\prime} & = \mathbf{x} +\dfrac{\left(\gamma-1\right)}{\beta^{2}}\left(\mathbf{x}\boldsymbol{\cdot}\boldsymbol{\beta} \right)\boldsymbol{\beta}-\gamma\boldsymbol{\beta}x_{0} \end{align} \Biggr\} \tag{11.19}$$


A Lorentz boost is just linear acceleration in a spatial direction.

An amazing fact that a lot of folks don't know: all linear acceleration in N dimensions is rotation in N + 1 dimensions.

Think of a candle in a dark room sitting on the edge of a rotating turntable. Look at it from the side. The candle appears to be accelerating, and then slowing down and reversing direction. This is in one dimension.

But seen from above, in 2 dimensions, a more complete view reveals simple rotation.

Linear acceleration in N dimensions is rotation in N + 1 -- and that's true in any number of dimensions.

Exercise for the student: what is an example in two and three dimensions?

So what does that have to do with the question?

Well, you can accelerate an object in the X direction, and its path is parallel and colinear to its previous path. This is accurate in 3 dimensions, but incomplete.

Seen as it really is, in 4 dimensions, the two world lines (4-momentum vectors) are NOT parallel.

The object that is moving faster is pointed slightly away from the direction of future time. Another name for this is time dilation. If the object accelerates all the way to c, its momentum vector would rotate 90 degrees, and it wouldn't be moving in the time direction at all.

The situation is symmetric, because the coordinate system is arbitrary. Each object will consider the other one as having veered off the "straight and narrow" path in which time passes at the normal rate -- i.e., the path that takes the least amount of proper time.

You're welcome! [curtsies]


If you understand that, but just barely, skip this part. But if you want to expend a little effort, there's a way to look at it that makes it much easier to see what's going on.

The reason this all seems so strange is that the universe is not euclidean. Momentum through time and space is contravariant. That means that when you increase your speed through one, you necessarily, always move slower through the other. This is a consequence of the fact that distance in one of the directions you can move (i.e, time) is negative relative to the other 3.

Space like this is called "pseudometric," and Einstein said that spacetime is like that.

This is his equation that defines the absolute (Lorentz-invariant 4-dimensional) distance between two events. Note that:

===[ Elapsed time is SUBTRACTED from the spatial distance between the events.

An easy way to look at it is to think of yourself as driving your car along the equator of a globe. Everyone moving at the same speed as you is also moving along the equator. The passage of time is motion in that direction.

People moving faster than you are also moving parallel to the equator, but they are on a parallel path slightly north of yours.

As they drive past you, you peek through their window and see that their clock is running slow.

If they suddenly accelerate to c, you would see them veer straight north. The North Pole is reachable at the speed of light.

Here's the thing:

When you accelerate, you're taking an exit on the highway that's just slightly more northward. But after you finish accelerating, you look out the window and you're amazed to see that you're still driving along the equator and the North Pole is still 90 degrees away from you.

This is actually what's really going on. The globe is a Riemann sphere, the North Pole represents infinite momentum, and the South Pole represents an object with zero mass-energy and zero momentum. Such an object would, by definition, not exist.

The Lorentz Transformation is just a map between Euclidean space and the surface of the Riemann sphere.

...That probably just makes it more confusing.

Oh well.

  • $\begingroup$ “that distance in one of the directions you can move (i.e, time) is negative relative to the other 3.” This should be “that distance squared in one of the directions you can move (i.e, time) is negative relative to the other 3.” $\endgroup$
    – Dale
    Commented 2 days ago
  • $\begingroup$ Yes of course. But making it more technically accurate makes it less easy to understand. If I were writing for publication, of course I would include the fact that the displacements are squared. But I take my lesson from Feynman, who never said "objects with mass" when talking to a popular audience. He said "stuff." $\endgroup$ Commented 2 days ago
  • $\begingroup$ Did you vote this down vote this because it's wrong, or because you don't understand it? Because if both, then it's the second one. And if you think it's the first, it's also the 2nd. Explain yourself. $\endgroup$ Commented 18 hours ago
  • 1
    $\begingroup$ I didn’t vote it down. I think it is ok except for the thing I pointed out in the comments, so that is enough for me. $\endgroup$
    – Dale
    Commented 17 hours ago
  • 1
    $\begingroup$ Ah, ok. The plural and singular “you” is an obnoxious thing about English. $\endgroup$
    – Dale
    Commented 17 hours ago

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.