In order to describe a notion of rate of change of positon, in four-dimensional spacetime, we have to introduce the concept of four-velocity.

So, consider the following:

For a massive particle with position $x^{\mu}(t) = (x^{0},x^{1},x^{2},x^{3}) \equiv (x^{0},\vec{x})$ we define the coodinate velocity as:$$ v^{\mu} := \frac{dx^{\mu}}{dt} \equiv (c,\vec{v})\tag{1}$$ Where the spatial components of $(1)$ coincide with classical velocity vector and t is the coordinate time.

But, $(1)$ is not a vector object indeed, because the components didn't transforms as vectors under a lorentz transformation:

$$\frac{dx'^{\mu}}{dt'} = \Lambda^{\mu'}_{\nu}\frac{dx^{\nu}}{dt} \frac{dt}{dt'} = \frac{\Lambda^{\mu'}_{\nu}}{\Lambda^{0'}_{\nu}x^{\nu}}\frac{dx^{\nu}}{dt} \neq \Lambda^{\mu'}_{\nu}\frac{dx^{\nu}}{dt}\tag{2}$$

Well, I simply do not understand one elementary derivation:

--> It's not difficult to know the motivation for this definition:

$$ v^{\mu} := \frac{dx^{\mu}}{dt} \equiv (c,\vec{v})$$

I mean, we need a four-vector and this is certainly a intuitive candidate, but then we realized that this object are not invariant under lorentz transformation, ok. But I'm struggling to derive the expression:

$$\frac{dx^{\mu}}{dt'} = \frac{dx^{\mu}}{dt} \frac{dt}{dt'} \tag{3}$$

Which is important to make the analysis of coordinate transformation as in $(2)$. I know that this is simply the chain rule structure, but I simply do not see how to derive it! The classical chain rule is then:

$$\frac{df[x(t),y(t),z(t)]}{dt} = \frac{\partial f }{\partial x }\frac{dx }{d t}+\frac{\partial f }{\partial y }\frac{d y}{d t}+\frac{\partial f }{\partial y }\frac{d y}{d t}$$

How can I derive $(3)$ from chain rule,explicitly?


2 Answers 2


You're misunderstanding the chain rule. For a function of a single argument, $$\frac{df}{dx} = \frac{df}{du} \frac{du}{dx}.$$ You only get multiple terms on the right-hand side if the function has multiple arguments. Now, the position $x^\mu(t)$ is simply a collection of four functions, each of one argument, $t$. So you just use the ordinary chain rule. For example, for the first function, $$\frac{dx^0}{dt} = \frac{dx^0}{dt'} \frac{dt'}{dt}.$$ The same holds for $dx^1/dt$, $dx^2/dt$, and $dx^3/dt$, so we conclude $$\frac{dx^\mu}{dt} = \frac{dx^\mu}{dt'} \frac{dt'}{dt}$$ for $\mu \in \{0, 1, 2, 3\}$.

  • $\begingroup$ Yes, I agree. As I understand the whole thing is derived as the following (in 1+1 dimensions; \mu = {0,1}): $$x^{\mu} = x^{\mu}(t'(t)) = [x^{0}(t'(t)), x^{1}(t'(t))] = $$ Then: $$\frac{dx^{\mu}}{dt}\vec{e}_{\mu} = \frac{dx^{0}}{dt}\vec{e}_{0} + \frac{dx^{1}}{dt}\vec{e}_{1} \implies $$ $$\frac{dx^{\mu}}{dt}\vec{e}_{\mu} = \frac{dx^{0}}{dt'}\frac{dt'}{dt}\vec{e}_{0} + \frac{dx^{1}}{dt'}\frac{dt'}{dt}\vec{e}_{1} \implies $$ $\endgroup$
    – M.N.Raia
    Commented Dec 1, 2018 at 21:53

When there are multiple variables in play, such as here $(t,x,y,z)$ it can be easy to get muddled on when we need partial derivatives, and when it is legitimate to give a total derivative. The central concept you need here is the concept of the worldline. This traces one particular line through spacetime. When we write a total derivative such as $dx^\mu /dt$ we are implicitly referring to this worldline. It is a statement about the difference $dx^\mu$ between adjacent events on the worldline. $dt$ is the coordinate time difference between those two events. It is only because we have this line, i.e. a one-dimensional or single-parameter entity, that total derivatives can be used. Once you know you have a single-parameter function, you can use expressions such as $$ \frac{df}{dr} = \frac{df}{ds} \frac{ds}{dr}$$ and thus, in your example, your eq (3).

But in the case of 4-velocity I think the reference you quoted is not going about it in the clearest way.

$dx^\mu$ is a 4-vector (an infinitesimal one) but $dt$ is not a Lorentz invariant scalar. To get a 4-vector you need to divide $dx^\mu$ by a Lorentz-invariant scalar. In the case of 4-velocity you divide $dx^\mu$ by the Lorentz invariant proper $d\tau$ time between the two adjacent events. Thus you get $$ \frac{dx^\mu}{d\tau}. $$


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.