From an older exercise sheet in my relativity class which I completed, but didn't annotate enough to remember it everything was correct:

Let $X$ be an inertial observer noticing that a spacetime event $B$ was caused by another event $A$. Show that no inertial observer $X'$ exists for whom the event $A$ is seen to be caused by $B$.

I chose the following ansatz:

Let $t_B>t_A$ in $X$. Then the Lorentz transformation of $t_A$ and $t_B$ gives us $$ t_A' = \frac{ct_A - \beta x_A}{\sqrt{1-\beta^2}} \quad t_B' = \frac{ct_B - \beta x_B}{\sqrt{1-\beta^2}} $$ Now, for causality to be given, we demand $0 < t_B - t_A$, i.e., the event $B$ happens after $A$. $$0 < t_B - t_A = \gamma (ct_B - \beta x_B - ct_A + \beta x_A)\\ =\gamma [c(t_B - t_A) - \beta (x_B - x_A)]$$ Losing the gamma and rearranging gives us $$ \beta (x_B - x_A) < c(t_B - t_A) \iff v/c^2 (x_B - x_A) < t_B - t_A $$ Now, since $v<c$, the expression $t_B - t_A \geq \frac{x_B - x_A}{c}$ holds.

I haven't written anything else after this, and I'm not quite sure whether a) this proves the causality-independence of the observer or b) my way of calculating is correct.

Looking forward to answers.


1 Answer 1


As you correctly note, you need to prove $t_B > t_A$ is preserved by Lorentz transformations. It's not clear to me whether your final answer demonstrates this (since the sign of $x_b - x_a$ isn't totally obvious), but I think you're on the right track.

I might have responded with the following argument. The invariant interval $$ds^2 = -dt^2 + dx^2 \tag1$$ is preserved by Lorentz transformations. Since event $B$ is "caused" by event $A$, $A$ must be in $B$'s past light cone, so the finite interval $s$ between them satisfies $$s^2 < 0 \tag2 .$$

Now (1) with $s$ held constant is the equation of a hyperbola. Thus, considering WLOG a reference frame centred on $B$, Lorentz transformations will shift $A$ along a hyperbola. From (2), that hyperbola is confined everywhere to $B$'s forward light cone, and so no observer exists for whom $t_A > t_B$. This completes the argument.

Note an almost identical argument can be constructed in in general relativistic spacetimes. It depends only on the fact that $t$ and $x$ have different signs in the interval.

  • $\begingroup$ I like your answer a lot, given it's argument on a geometric basis. $\endgroup$
    – John Doe
    Commented Nov 21, 2015 at 13:48

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.