There's several ways to look at this. The classical way is picture the photon climbing out of the deeper well gravitational potential of the denser early Universe. What that suggests is in fact the photon finds itself being measured in a reference frame difference from the one it was emitted in. In fact from the photon's point of view it hasn't interacted with anything and has the same energy it started off with. So if you rotate your reference frame to be the same as when it emitted (ie fly towards it and blue shift it back up) it looks as it has the same energy.) So there's no quantum interaction necessary it's just relativity and coordinate frames.

Unfortunately energy is not conserved in general relativity as we understand it in Newtonian terms. There's a blog post by Sean Carroll on this.

http://www.preposterousuniverse.com/blog/2010/02/22/energy-is-not-conserved/

When photons are subject to the cosmological redshift?

What Robert Anthony Watson said is commonly called as the Sachs–Wolfe effect feature in CMBR spectrum. You can find more details here..

https://en.wikipedia.org/wiki/Sachs%E2%80%93Wolfe_effect

http://ifa.hawaii.edu/cosmowave/supervoids/the-integrated-sachs-wolfe-effect/

What happens to photon energy as the universe expands?

The rest mass energy of a nonrelativistic particle (such as, Baryons) remains constant in time, but since the number density (ρ) of such particles is inversely proportional to the volume of space due to expansion, they have the scaling, ρ ∝ a^-3, where a is the scale factor.

On the contrary, for radiation and particles like neutrinos which have velocities very close to the light speed, there is an extra degree of freedom of scaling in time; in addition to the usual spatial scaling. Energy of radiation and relativistic ultra-hot matter (neutrinos) is inversely proportional to its wavelength as well i.e., h nu ∝ a^-1 due to the cosmological redshift, thus the total scaling in this case is ρ ∝ a^-4. Hope it helps!

Dear Francis,

I am also adding a presentation slide here in addition to my comments above. What you have asked is quite technical, however the slide here will provide the right intuition. The slides were a part of the lecture notes of course AST5220 at University of Oslo. I am not sure, if MS office can display the animation correctly. You can try it, otherwise LibreOffice or OpenOffice would do.

Dr. Watson,

I don't understand you reference to the photon's inertial frame. As i learned SP theory, as a frame of reference approaches the speed of light, it's proper time goes to zero. In other words a photon ceases to exist at the same time it is created in its reference frame. The only way i can think of photon interaction, therefore, is if a new photon is created during, for example, Compton scattering. So, a photon doesn't know it has interacted because it ceases to exist at the interaction. I did read the blog you referenced. Very interesting. For the moment i'm holding on to the idea that there is an interaction between photons and gravitons as the universe expands, and that energy is actually stored in a quantum gravitational field. This seems to coincide with the Newtonian limit of weak fields (minus the quantum). I'm not quite ready to give up on energy conservation.

Hi Francis,

Sorry I wasn't referring to the photon's reference frame which I think is undefined, but to the co-moving reference frame of the plasma it was emitted from.  Indeed if we knew exactly how space-time works at the quantum level then all this would drop out.

Thanks for your input, Dr. Watson. I agree it's hard to think of a reference frame whose proper lifespan is zero. I'd like your idea (or anyone's) on this followup question. It has been my thought that the gravitational Doppler effect, which you mentioned, may be the reason for the cosmological redshift. One thing that bothered me about the expansion of the universe is why galaxies get farther from each other but no expansion seems to be evident within galaxies. And why don't meter sticks get longer as the universe expands. One possible explanation (probably naive) that occurred to me that might at least apply to meter sticks is that virtual photons, like real ones, may be subject to gravitational redshift. Except, unlike "cosmic" photons, they don't exist long enough in a frame of reference at rest with a meter stick to experience such a redshift. That might explain why meter sticks don't get longer, but of course galaxies are held together by gravity, not EM virtual quanta. Maybe the cosmic redshift can affect graviton interaction between galaxies but not so much between parts of galaxies? The idea  that the gravitational Doppler shift might explain the cosmic redshift has led me to think this might require the existence of the graviton, since the photon would have to have some quantum interaction with the field. Before that thought i didn't like the idea of the graviton.

Francis,

      The over dense regions expand slower than the under denser parts of the Universe, and in the end decouple from the Universal expansion due their own self-gravity (due mainly to cold dark matter). Within these over-dense halos baryons can form galaxies which are, yes, balanced by gravity and rotational forces. So there's nothing really special going on, it's just the metric covering the local group of galaxies in a cluster is dominated by the stable CDM halo which is different to the Universal expansion metric.   

Thanks, that makes a lot of sense. So, the cosmic "balloon" shouldn't expand equally. There are hills and valleys. But i still wonder if something like a graviton, should it exist, is subject to space expansion. And what about virtual photons?

That's right. Early on the bits of evolve in isolation based on local density as if the whole Universe was at that density. The low density parts overtake them and become these 100 Mpc voids inside a web-like network. So it's more like a bursting 4D balloon. For the graviton and virtual photons you need a tame theoretic physicist, but my gut feelings it would be virtual photon would be, but gravitons wouldn't since somehow they are part of the process.

 The Pearlman Spiral offers an alternate hypothesis of Cosmological Redshift, consistent with a static universe, not cosmological expansion. assuming valid, would that render the question moot?

Formulated in April of 2013, first public review edition just out, those able please review and recommend to the extent warranted, Thank you,

roger m.

http://www.amazon.com/dp/1517214122/ref=cm_sw_r_fa_dp_ijX8vb1W3YWZ9