The four-momentum in special relativity is defined as: $$p = \bigg(\frac{E}{c},\ p_x,\ p_y,\ p_z\bigg)$$ where $$p_x = \gamma (mv_x)$$ $$p_y = \gamma (mv_y)$$ $$p_z = \gamma (mv_z)$$

When solving problems, it is assumed that the individual components of this four-vector at any time will have a constant value. In other words, we assume that this quantity is conserved. How can we prove this fact?


Consider the action of a free relativistic particle of mass $m$, given by,

$$S = -m\int dt \, \sqrt{1-\dot x^2}$$

and we can find the Euler-Lagrange equations imply,

$$\frac{\partial}{\partial t}\frac{\partial L}{\partial \dot x} = \frac{\partial}{\partial t} \frac{m\dot x}{\sqrt{1-\dot x^2}} = 0$$

since $\partial L/\partial x = 0$. Note that we can identify $\partial L/\partial \dot x = p = \gamma m\dot x$. Thus, the Euler-Lagrange equations are the statement that the relativistic momenta are conserved. The Hamiltonian is given by a Legendre transform,

$$H = p\dot x - L = \sqrt{m^2 + p^2}$$

and one can see that,

$$\frac{\partial H}{\partial t} = \frac{p \dot p}{\sqrt{m^2+p^2}} = 0$$

since $\dot p =0$ by the Euler-Lagrange equations implying conservation of momentum. We thus have that the energy and relativistic momenta are conserved and so $p^\mu$ itself is conserved.

Both of these conservation laws are a consequence of the fact the system is invariant under translations in time, and the action does not depend on the position of the particle, since only its derivative $\dot x$ appears in the action. By Noether's theorem, this implies energy and momentum conservation.

  • $\begingroup$ Is there any way to show the conservation law without the use of Lagrangian Mechanics? I am not yet familiar with it. $\endgroup$
    – Anonymous
    Jul 10 '17 at 16:49
  • $\begingroup$ Playing devil's advocate here: how do you know it's the right action without invoking that it gives the right answer? $\endgroup$ Jul 10 '17 at 17:08
  • $\begingroup$ @ZeroTheHero It's analogous to the Nambu-Goto action for the relativistic string that sweeps out a worldsheet, except for a particle it sweeps out a wordline. Moreover, the action written in a different form, is equivalent to the one that gives rise to the geodesic equations in general relativity, that is, $\sqrt{g_{\mu\nu}\dot x^\mu \dot x^\nu}$. $\endgroup$
    – JamalS
    Jul 10 '17 at 17:18
  • $\begingroup$ @JamalS I don't want to belabour the point too much because I know you have the right action. My observation is that the starting point of this is the experimental observation that the 4-momentum as given in the OP's question is conserved. Whatever action you come up with has to include this conservation law as a consequence. You can come up with all the elegant arguments in the world as to why this or that action should be right, if it is doesn't include conservation of 4-momentum, it's not going anywhere. $\endgroup$ Jul 10 '17 at 17:24
  • $\begingroup$ @ZeroTheHero I think an argument can be made analogous to bubbles, in that they minimise their surface area with respect to the volume they enclose. Since the action is a one-dimensional version of such an action, maybe there is some way to make a similar argument, that isn't self-referential with respect to the conservation of momentum. $\endgroup$
    – JamalS
    Jul 10 '17 at 17:38

If a Lagrangian has the symmetry to be translationally invariant, by Noether's theorem, you can show the corresponding conserved quantities are energy and momentum.


There is no a priory way of showing that 4-momentum as defined is conserved. It is however a reasonable guess and, more importantly, it is experimentally verified that this definition leads to conservation of 4-momentum.

I can't for the life of me immediately think of another reasonable definition - it's a 4-vector, it reduces to the usual conservation in the limit of small velocities, etc - but if this definition had failed the experimental test someone would have come up with another definition compatible with experiment.

  • $\begingroup$ But JamalS has provided a justification for it using the techniques of Lagrangian mechanics. What is wrong with that? $\endgroup$
    – Anonymous
    Jul 10 '17 at 17:06
  • $\begingroup$ @Anonymous see my comment to Jamal's answer. (Not suggesting JamalS is wrong in anyway.) $\endgroup$ Jul 10 '17 at 17:09
  • $\begingroup$ So, it's just an experimental fact and not a consequence of the postulates? $\endgroup$
    – Anonymous
    Jul 10 '17 at 17:59
  • $\begingroup$ @Anonymous As per my last comment to JamalS, AFAIK it is primarily an experimental fact. The form of the 4-vector is dictated by the requirement of properly transforming under Lorentz - that's theoretical. Of course as mentioned above it is immediately included in any theory, and - I want to emphasize this - I don't know of any other reasonable alternative that would compete with the current form. Basically what theory shows is that, if conservation of 4- momentum as defined in your question holds in one inertial frame, it must hold in any other Lorentz-connected frame. $\endgroup$ Jul 10 '17 at 18:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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