Is Modified Gravity Theory a Viable Alternative to Dark Matter?

[twofish-quant]

We don't just add whatever dark matter we need to balance the sums, instead you take the predictions from simulations which have modeled the gravitational evolution of structure from the early universe to today (or to whatever redshift you are looking at) and compared that to observations. There is not the freedom to invent whatever mass is needed, you're constrained by the physics, which gives results which agree with observations.

And the other thing is that there are *tons* of observations. One other thing is that a lot of the observations are culmulative, which means that my just taking more of the same type of observation, you beat down the errors, and you have situations in which knowing that the number is 2, doesn't make a difference but knowing that the numbers if 2.3 and not 2.6 makes a huge difference.

To be fair, the current data is possibly not good enough, the extra model complexity in something like TeVeS means you prefer a simpler model like LCDM unless the data was clear enough.

One reason that people also tend to dark matter is that it's easier to intuitively think about dark matter than general relativity. General relativity (and generalizations of it like TEVES) is a beautiful elegant theory. It's also a pain in the rear end to get any sort of calculation from it. So if you do a calculation, you *assume* that weird gravity isn't that important, because if it's not important, you get a result. If your results don't make sense then you bite the bullet and go back and add in the complex stuff later.

This is the sort of thing that leads to scientific revolutions. If there is something from measurements that suggests that TEVES explains some weird thing about the early universe, then what I'll do is to spend about three months and put TEVES physics into my supernova code. Now it might be that when I do that, suddenly I get realistic explosions, in which case I publish something, which causes people to apply TEVES to their stuff, and you could get a snowball effect in which within a year or two, modified gravity becomes the new "standard model"

Or maybe not. The reason I'm not putting strange gravity into my supernova code right now is because it's going to take three to six months to get it to work, and *right now* there are other pieces of physics that I could spend my time looking at. (And yes, people have tried to put dark matter candidates into supernova code, and that doesn't do much.)

 

[twofish-quant]

Arguably the biggest knock against many (not all) of the modifications to GR is that they are inconsistent at the quantum level. They have ghost modes that do not cancel.

On the other hand straight GR is not renormalizable at the quantum level. I've heard it said that it's a good thing that Einstein proposed GR when quantum field theory was less will understood, because people would have used the non-renormaliziability of GR as evidence against it.

So while the quantization of gravity is a big mystery in general, it of course has to happen at some point and any theory that fails the most basic principles (loss of unitarity, gauge anomalies, etc) should be ruled out.

Observation beats theory. If it turns out that we need to modify GR to get cosmology to work, this is just more funding for theorests to think about the problem. Also pretty much all of the major cosmological theories are phenomenonlogical. We have this fudge factor parameter that we adjust to make everything fit. What's fun about this is that it turns out to be really, really hard to get that to work.

 

[twofish-quant]

This is also an example of how "real science" doesn't match "science in the movies." There is this idea of this lone genius that fights the "establishment" and then wins from shear brilliance. It doesn't really work that way.

 

[atyy]

On the other hand straight GR is not renormalizable at the quantum level. I've heard it said that it's a good thing that Einstein proposed GR when quantum field theory was less will understood, because people would have used the non-renormaliziability of GR as evidence against it.

But renormalizability isn't a requirement for a useful quantum field theory, so GR is a useful QFT.

 

[Haelfix]

GR may or may not be a useful quantum theory (it might have to be modified at very high energies), otoh it is not a manifestly inconsistent quantum theory, at least at this level of discussion.

A theory with residual gauge or conformal anomalies, or those with ghosts by contrast is always inconsistent, and is thus ruled out automatically from the beginning. For instance, Palatini f(R) gravity is an example, as well as many of the higher derivative theories, bimetric ones and so forth. You would have to insist upon a modification of quantum mechanics to get them to work.

Its a very nice constraint, b/c some of those theories can be tweaked to pass purely cosmological and relativity tests (eg the absense of curvature singularities, no ctcs, etc).

 

[atyy]

GR may or may not be a useful quantum theory (it might have to be modified at very high energies), otoh it is not a manifestly inconsistent quantum theory, at least at this level of discussion.

I was just thinking along the lines of Donoghue's GR as an effective quantum field theory at low energies, needing a still unknown UV completion.

 

[twofish-quant]

For instance, Palatini f(R) gravity is an example, as well as many of the higher derivative theories, bimetric ones and so forth. You would have to insist upon a modification of quantum mechanics to get them to work.

So modify quantum mechanics. I don't have much of a problem with that.