No. No. Not THAT controversy. I just got back from a meeting at the insanely cool Carnegie Observatory in Pasadena, California (Hubble's old digs) with Wendy Parker, Paul Humphreys, James Ladyman, and many extremely interesting and engaging astronomers. (Barry Madore and Wendy Freedman were our hosts, and Stacy McGaugh, Bill Saslaw, Alar Toomre (who did an insanely cool computer simulation of galaxy collision back in 1972!), Frank van den Bosch, and James Bullock were in attendance.) Among the many interesting things I learned is that the whole issue of Dark Matter is much more complicated than I ever imagined. I used to think that Dark Matter was simply sprinkled liberally around the universe in exactly the right quantities to get the accelerations of galaxies and clusters right. I thought, in other words, that it was a simple and straightforward Duhem problem. Then, a few years ago, the images of the Bullet supercluster collision came out, and it was widely reputed to offer direct evidence of dark matter. This made it seem like a standard "independant confirmation" of a Duhumian auxilliary hypothesis. But of course, the situation is MUCH more complicated. There are in fact at least 16 different moving parts in the Dark Matter controversy, and superclusters are just one of them. And in fact, long before the Bullet cluster was observed, opponents of the standard model of "cold dark matter" (who advocate a modified theory of gravity) had admitted that superclusters probably had missing mass. But the rub is this: there is plenty of known missing BARYONIC mass (the kind that the Big Bang neucleosynthesis model predicts the expected quantity of) to account for the missing mass in superclusters. In fact, that would only take about 3% of the baryonic mass out there, and as much as 30% is known to be missing. So, the "dark matter" in the Bullet supercluster could easily be brown dwarfs, black holes, or other "normal" stuff. So, the debate is much more intricate, and it involves trying to figure out how the dark matter halos of the universe would have evolved from the tiny fluctuations of the cosmic microwave background and then predicting what velocity curves for galaxies those would produce. The contest is then to see who can predict velocity curves better: the people with modified theories of gravity, or the people who simulate the dark matter and then see what it does. From what I can tell, the modified gravity people seem to have the edge in that, and they do with fewer free parameters. All very fascinating stuff.
4 responses to “Teach the Controversy”
-
“the modified gravity people seem to have the edge in [predicting velocity curves], and they do with fewer free parameters.”
Can you elaborate on that? Of course, the issue isn’t number of free parameters, but how tightly the theory constrains the phenomena. But, having heard Stacy McGaugh talk about this in Michigan a few years ago, I got the impression that TeVeS, the best candidate for a modified gravitation theory, was not so much a theory as a family of theories, with, not just free parameters, but a free function f on which there were only weak theoretical constraints.
Does the theory, for all choices of the free function, constrain the rotation curves? If not, can we get information about that function from other phenomena?LikeLike
-
Hi Wayne,
TeVeS is a family of relativistic theories. Stacey only talked about the non-relativistic version called MOdified Newtonian Dynamics. It constrains the phenomena pretty tightly–tightly enough to have made some impressive predictions. MOND basically has one parameter, a value of acceleration below which any test particle acted on by a gravitational force gets an extra boost. (or at least that’s how I understood it). Milgrom set all this long before many of the data came in, and it gets a lot of the data right. The thing about the dark matter predictions is that they all involve simulations of dark matter clumping from the CMB and those simulations all involve sub-grid “physics”–which have tons of parameters. In those, the phenomena seem to be mostly constrained, if at all, by tradition, and by the fact that the practitioners only tune to the data so often. And they still don’t predict very well. In some sense it seems absurd to say that theory B constrains the phenomena far less AND doesn’t come as close to matching them, but that does seem to be what’s going on simply because they have adapted slowly.
In neither case, (MOND or Dark matter) is there as much constraint on the theory by outside phenomena as we would like. A comment I made at the meeting is that we were doomed to epicycles until Kepler and Newton started unifying astronomy and terrestrial physics. As long as there is no terrestrial physics of either dark matter or phenomena that would be comparable to the galactic regime, we may be doomed to abiding more epicycles.LikeLike
-
MOND got gravitational lensing wrong—predicted not enough, which is why most people thought that the whole program was dead in the water. That was one of the breakthroughs of TeVeS.
As I understand it (and it’s been a while since I’ve looked into these things), gravitational lensing gives us an estimate of the mass of the dark matter halos around galaxies that agrees with the estimates we get from rotation curves. If that’s right, it’s something that we should take seriously—on theories with no dark matter, are these agreeing estimates mere coincidence?LikeLike
-
OK, I’m stretching beyond my limited comprehension here, but:
1)you need to get more than the mass of the halos right, you need to get the shape right, cusps vs cores and all that. So, you want to get the lensing right, which I think only depends (according to GR) on total mass, and you want to get the rotation curves right, which depends on the specific arrangement of the dark matter. The interesting thing is that the dark matter model would lead you to think you should have a really hard time predicting the acceleration curves from only the luminous matter, but with the exception of super clusters (which make up a tiny fraction of the mass, and hence could be affected substantially by missing baryonic mass) you can do it very easily. Now THAT is a coincidence which looks peculiar from the point of view of CDM.
2)I thought TeVeS was just a generalization of MoND for getting the lensing right. If it does that right, then the coincidence of CDM getting a match between lensing and acceleration is, I take it, explained. But now we are back to your question of how much flexibility TeVeS has and I’m out of my league. I’m guessing it goes like this: you use a simple MoND model to predict (almost) all the curves right just from the luminous matter. you then use Beckenstein’s method to adapt that MoND model so that it gets the lensing right. But I’m extremely shaky on two crucial issues: a)what’s the relationship between Milgrom’s simple MoND model and the TeVeS generalization one needs to produce in order to get lensing. and b)does the level of flexibility that is available to you when you do what’s involved in a) trump the level of tuning that the CDM people need to do to get cores instead of cusps, etc, when they try to predict the curves. All I want to emphasize here is that there are more moving parts than I think most people realize. (And I realize that you, Wayne, are not “most” people!)
Something else I didnt mention: CDM is terrible at getting the smallest satellite galaxies right, even with all the tuning available. This is leading some (like James Bullock–who was at the meeting) to postulate a self interaction term for the DM. Nothing wrong with that, in principle, of course. But the ante on loosening the constraints on phenomena just got upped.LikeLike
New APPS 
Leave a comment