2

I'm currently reading Gravitation by MTW, and in it they postulate that the energy of a photon in a moving observers reference frame is, $$E = -\mathbf{p} \cdot \mathbf{u}$$ Where $\mathbf{p}$ is the 4-momentum and $\mathbf{u}$ is the 4-velocity of the observer.

How would I prove the above and what is the intuition that it is true? I know that for a photon in dimensionless units is, $$E = h/\lambda$$ My guess is that we can look at this as a phenomenon due to a length contraction with respect to the moving observers frame which results in a different observed wavelength of the photon.

Qmechanic
  • 201,751
Jeff
  • 221

1 Answers1

4

In the $(-,+,+,+)$- signature, "$ -\hat u\cdot$" is the operation "take the time-component along the 4-velocity $\hat u$ [that is, "according to the observer along $\hat u$].

It applies to any 4-vector. [The Euclidean analogue is "$\hat x\cdot$" is the operation "take the x-component of a vector".]

For any 4-momentum $\tilde p$ (for a photon or a particle),
$$ -\hat u\cdot \tilde p= E$$ is the relativistic energy of the particle according the observer with 4-velocity $\hat u$. So, it defines the "relativistic energy".

(By the way, "length contraction" (involving the spatial separation between two parallel timelike worldlines) doesn't have anything to do with "observed wavelength" (involving the spatial separation between two lightlike lines). Note that length contraction is independent of direction (forward or backward) of relative-motion, but doppler effect depends on direction.)
See: Deriving Relativistic Doppler Effect through length contraction


UPDATE:

Consider another observer with 4-velocity $\hat v$.

Our observer $\hat u$ would write: $$\hat v=\gamma \hat u + \gamma V \hat u_{\perp}$$ $$\hat v_{\perp}=\gamma \hat V u + \gamma \hat u_{\perp}$$ where $\hat u\cdot\hat u_{\perp}=0$ and $\hat v\cdot\hat v_{\perp}=0$ and all vectors here are coplanar.

Let the 4-momentum of particle at rest in $\hat v$ be the timelike vector $$\tilde P=m\hat v$$ and let the 4-momentum of a light-signal $\tilde K$ be expressed by $\hat v$ as $$\tilde K=k\hat v + k\hat v_{\perp}$$ since $\tilde K$ is lightlike: $\tilde K \cdot \tilde K=0$.

In terms of $\hat u$ and $\hat u_{\perp}$, we have $$\tilde P=m\hat v= m\left( \gamma \hat u + \gamma V \hat u_{\perp} \right) $$ so, that $\hat u$ measures a time-component of $\tilde P$: $$-\hat u\cdot \tilde P =-\hat u\cdot \left( m\left( \gamma \hat u + \gamma V \hat u_{\perp} \right) \right)=\gamma m,$$ which we recognize as the relativistic energy of a massive particle.

In terms of $\hat u$ and $\hat u_{\perp}$, we have $$\begin{align} \tilde K=k \hat v + k\hat v_{\perp} &= k\left( \gamma \hat u + \gamma V \hat u_{\perp} \right)+ k\left( \gamma V \hat u + \gamma \hat u_{\perp} \right)\\ &= k\left( \gamma +\gamma V \right)\hat u+ k\left( \gamma V + \gamma \right)\hat u_{\perp} \\ &= k\gamma(1+ V)\left( \hat u+ \hat u_{\perp}\right) \\ \end{align} $$ so, that $\hat u$ measures a time-component of $\tilde K$: $$-\hat u\cdot \tilde K =-\hat u\cdot \left( k\gamma(1+ V)\left( \hat u+ \hat u_{\perp}\right) \right) =k\gamma(1+ V). $$ Note that $\gamma(1+V)=\frac{1}{\sqrt{1-V^2}}(1+V) =\frac{\sqrt{(1+V)(1+V)}}{\sqrt{(1+V)(1-V)}} =\sqrt{\frac{1+V}{1-V}} $ is the Doppler factor. So, we could associate the temporal component of $\tilde K$ as the relativistic energy of a light signal.

robphy
  • 11,748
  • When you said "take the time-like component", are you referring to the time component? I always thought that a time-like vector is anything inside of the light cone, not a specific direction. What is the time-like component for a light-like photon? – Jeff Dec 25 '20 at 21:45
  • I probably should have said take the "time-component according to $\hat u$". (I'll correct it. Thanks.) – robphy Dec 25 '20 at 21:51
  • I'm a bit hung up on your comment that $E = -\mathbf{p} \cdot \mathbf{u}$ defines relativistic energy for any particle. It seems like that would be easy to show for ordinary particles from the Lorentz transformations for velocities. However, the velocity of light is the same in all frames, so I don't quite get where this is derived from in that case. It seems that instead we want to derive the doppler shifts as you described in the other question and then use those to see how the energy of a photon changes using $E = h/ \lambda$. Which, I assume works out to the same relation. – Jeff Dec 25 '20 at 22:19
  • @Jeff I updated my answer. – robphy Dec 26 '20 at 01:02
  • Thanks for working this out. I am still working through some of the details, but it looks like it satisfies my answer. – Jeff Dec 26 '20 at 03:57
  • I am struggling with one part of your update. You said let the 4-momentum of a light signal be expressed as $K = k v + k v_{\perp}$. Did you specifically choose the 4-momentum to be in that direction for simplicity? It seems that an arbitrary light signal could be anywhere on the light cone and not necessarily equal to that expression. – Jeff Dec 26 '20 at 04:09
  • @Jeff Yes, it is for simplicity. Everything is taking place in the "t-x plane". – robphy Dec 26 '20 at 05:38