The TA blasted through a problem from Hartle , section 5.17 (all the while apologizing for going so slow). I’m going to have to look these notes over carefully to figure out what on Earth he was doing.
At one point he asked if anybody was completely lost. Nobody said yes, but given the class title, I had the urge to say “No, just relatively lost”.
In a source’s rest frame emits radiation isotropically with a frequency with number flux . Moves along x’-axis with speed in an observer frame (). What does the energy flux in look like?
A brief intro with four vectors
For this we have the dot product
Greek letters in this course (opposite to everybody else in the world, because of Landau and Lifshitz) run from 1 to 3, whereas roman letters run through the set .
We want to put space and time on an equal footing and form the composite quantity (four vector)
It will also be convenient to drop indexes when referring to all the components of a four vector and we will use lower or upper case non-bold letters to represent such four vectors. For example
Three vectors will be represented as letters with over arrows or (in text) bold face .
Recall that the squared spacetime interval between two events and is defined as
In particular, with one of these zero, we have an operator which takes a single four vector and spits out a scalar, measuring a “distance” from the origin
This motivates the introduction of a dot product for our four vector space.
Utilizing the spacetime dot product of 1.13 we have for the dot product of the difference between two events
From this, assuming our dot product 1.13 is both linear and symmetric, we have for any pair of spacetime events
How do our four vectors transform? This is really just a notational issue, since this has already been discussed. In this new notation we have
where , and .
In order to put some structure to this, it can be helpful to express this dot product as a quadratic form. We write
We can write our Lorentz boost as a matrix
so that the dot product between two transformed four vectors takes the form
Back to the problem.
We will work in momentum space, where we have
Now, the TA blurted all this out. We know some of it from the QM context, and if we’ve been reading ahead know a bit of this from our text  (the energy momentum four vector relationships). Let’s go back to the classical electromagnetism and recall what we know about the relation of frequency and wave numbers for continuous fields. We want solutions to Maxwell’s equation in vacuum and can show that such solution also implies that our fields obey a wave equation
where is one of or . We have other constraints imposed on the solutions by Maxwell’s equations, but require that they at least obey 1.28 in addition to these constraints.
With application of a spatial Fourier transformation of the wave equation, we find that our solution takes the form
If one takes this as a given and applies the wave equation operator to this as a test solution, one finds without doing the Fourier transform work that we also have a constraint. That is
So even in the continuous field domain, we have a relationship between frequency and wave number. We see that this also happens to have the form of a lightlike spacetime interval
Also recall that the photoelectric effect imposes an experimental constraint on photon energy, where we have
Therefore if we impose a mechanics like relativistic energy-momentum relationship on light, it then makes sense to form a nilpotent (lightlike) four vector for our photon energy. This combines our special relativistic expectations, with the constraints on the fields imposed by classical electromagnetism. We can then write for the photon four momentum
Back to the TA’s formula blitz.
Utilizing spherical polar coordinates in momentum (wave number) space, measuring the polar angle from the (x-like) axis, we can compute this polar angle in both pairs of frames,
Note that this requires us to assume that wave number four vectors transform in the same fashion as regular mechanical position and momentum four vectors. Also note that we have the primed frame moving negatively along the x-axis, instead of the usual positive origin shift. The question is vague enough to allow this since it only requires motion.
as (ie: our primed frame velocity approaches the speed of light relative to the rest frame), , . The surface gets more and more compressed.
In the original reference frame the radiation was isotropic. In the new frame how does it change with respect to the angle? This is really a question to find this number flux rate
In our rest frame the total number of photons traveling through the surface in a given interval of time is
Here we utilize the spherical solid angle , and integrate over the interval. We also have to assume that our number flux density is not a function of horizontal angle in the rest frame.
In the moving frame we similarly have
and we again have had to assume that our transformed number flux density is not a function of the horizontal angle . This seems like a reasonable move since and as they are perpendicular to the boost direction.
Now, utilizing a conservation of mass argument, we can argue that . Regardless of the motion of the frame, the same number of particles move through the surface. Taking ratios, and examining an infinitesimal time interval, and the associated flux through a small patch, we have
Part of the statement above was a do-it-yourself. First recall that , so evaluated at is .
The rest is messier. We can calculate the values in the ratio above using 1.34. For example, for we have
If one does the same thing for , after a whole whack of messy algebra one finds that the differential terms and a whole lot more mystically cancels, leaving just
A bit more reduction with reference back to 1.34 verifies 1.41.
Also note that again from 1.34 we have
and rearranging this for gives us
which we can sum to find that
so putting all the pieces together we have
The question asks for the energy flux density. We get this by multiplying the number density by the frequency of the light in question. This is, as a function of the polar angle, in each of the frames.
But we have
The TA then writes
although, I calculate
He then says, the forward backward ratio is
The forward radiation is much bigger than the backwards radiation.
For this I get:
It is still bigger for positive, which I think is the point.
If I can somehow manage to keep my signs right as I do this course I may survive. Why did he pick a positive sign way back in 1.34?
 J.B. Hartle and T. Dray. Gravity: an introduction to Einsteins general relativity, volume 71. 2003.
 L.D. Landau and E.M. Lifshits. The classical theory of fields. Butterworth-Heinemann, 1980.