Milky Way Satellites and Dark Matter

I found a strange paper on the ArXiv last week, and was interested to see that it had been deemed to merit a press release from the Royal Astronomical Society that had been picked up by various sites across the interwebs.

The paper, to appear in due course in Monthly Notices of the Royal Astronomical Society, describes a study of the positions and velocities of small satellite galaxies and other object around the Milky Way, which suggest the existence of a flattened structure orientated at right angles to the Galactic plane. They call this the “Vast Polar Structure”. There’s even a nifty video showing this arrangement:

They argue that this is is evidence that these structures have a tidal origin, having been thrown out   in the collision between two smaller galaxies during the formation of the Milky Way. One would naively expect a much more isotropic distribution of material around our Galaxy if matter had fallen into it in the relatively quiescent way envisaged by more standard theoretical models.

Definitely Quite Interesting.

However, I was rather taken aback by this quotation by one of the authors, Pavel Kroupa, which ends the press release.

Our model appears to rule out the presence of dark matter in the universe, threatening a central pillar of current cosmological theory. We see this as the beginning of a paradigm shift, one that will ultimately lead us to a new understanding of the universe we inhabit.

Hang on a minute!

One would infer from this rather bold statement that the paper concerned contained a systematic comparison between the observations – allowing for selection effects, such as incomplete sky coverage – and detailed theoretical calculations of what is predicted in the standard theory of galaxy formation involving dark matter.

But it doesn’t.

What it does contain is a simple statistical calculation of the probability that the observed distribution of satellite galaxies would have arisen in an exactly isotropic distribution function, which they conclude to be around 0.2 per cent.

However, we already know that galaxies like the Milky Way are not exactly isotropic, so this isn’t really a test of the dark matter hypothesis. It’s a test of an idealised unrealistic model. And even if it were a more general test of the dark matter hypothesis, the probability of this hypothesis being correct is not what has been calculated. The probability of a model given the data is not the same as the probability of the data given the model. To get that you need Bayes’ theorem.

What needs to be done is to calculate the degree of anisotropy expected in the dark matter theory and in the tidal theory and then do a proper (i.e. Bayesian) comparison with the observations to see which model gives the better account of the data. This is not any easy thing to do because it necessitates doing detailed dynamical calculations at very high resolution of what galaxy like the Milky Way should look like according to both theories.

Until that’s done, these observations by no means “rule out” the dark matter theory.

36 Responses to “Milky Way Satellites and Dark Matter”

  1. “The probability of a model given the data is not the same as the probability of the data given the model. To get that you need Bayes’ theorem.”

    Or just common sense. Data: person is pregnant. Model (hypothesis): person is female. Probability of model given the data: 100%. Probability of data given the model: about 2%.

    “Our model appears to rule out the presence of dark matter in the universe”

    At best, without further assumptions, it could rule out dark matter in a certain part of our galaxy.

  2. Bryn Jones Says:

    I saw the RAS press release and then looked the preprint of the Pawlowski, Pflamm-Altenburg and Kroupa paper. I skim read it.

    The flattened distribution of galaxies in the vicinity of the Galaxy has been known for a long time, of course. This does appear to be real, even though there are observational biases. Among these biases is the Milky Way obscuring galaxies behind it. Many of the more recent discoveries of Local Group galaxies have come from the Sloan Digital Sky Survey which has sampled only parts of the sky. Another consideration is that Local Group dwarf galaxies appear to be clumped around the two massive spiral galaxies – around our own Galaxy and around M31.

    The Pawlowski, Pflamm-Altenburg and Kroupa paper performs some interesting analyses on tidal streams. The consideration of the likelihood that Local Group dwarf galaxies might have a tidal origin is useful.

    However, I agree with Peter that to test whether the observed distribution of Local Group galaxies is consistent with what would be expected under a standard cold dark matter scenario would require a detailed numerical comparison with data from N-body simulations, and with high-resolution simulations in particular. It’s not clear heuristically to me that the observed distribution differs from the structures that we might expect to find in the filaments within the galaxy distribution predicted in N-body studies. I’m therefore very reluctant to jump from the observed distribution to the conclusion that there are problems with the standard cold dark matter scenario with general relativity.

    The other point of the paper – arguing that Local Group dwarf galaxies were produced in an ancient tidal interaction, and then pointing out that those dwarfs would not be the low mass systems (haloes) predicted from cold dark matter simulations – is more interesting. However, I feel more work is needed to establish whether the dwarfs were formed in a single tidal interaction.

    I’m puzzled that the RAS hyped the work to the maximum extent. The paper is interesting and important, but the RAS press release did seem to go too far.

  3. John Hagel Says:

    Check out these very recent papers by the same team:
    “Can filamentary accretion explain the orbital poles of the Milky Way satellites?”
    and the major (long) review
    “The dark matter crisis: falsification of the current standard model of cosmology”

    So it seems the team has looked in much detail at all the possibilities, including the wider picture. This then perhaps allows one to understand Kroupa’s statement and the contents of the abstract.

    • telescoper Says:

      The first paper is quite interesting. Thanks for pointing it out. It still doesn’t deal with the selection effects in the sample described in the blog post above, nor does it do a full model comparison (accounting for resolution effects)…

      • John Hagel Says:

        The selection effects are dealt with (they are trivial, since, as they point out, only the 2pi obscuration near the disk of the Milky Way plays a role). Resolution issues: the cosmological models being compared with are the state of the art, i.e. the highest resolution computed in order to work out the arrangement of the dark matter sub halos in a large number of computer simulations. So the results are really significant. My impression is that one would need to resort to magic in order to fit the VPOS to the dark matter models.

      • telescoper Says:

        They’re not that trivial to deal with. Some of the objects discussed are identified in the SDSS survey which does not cover the whole sky, even out of the Galactic plane…

      • In a talk about a year ago, Eva Grebel mentioned this by now well known problem (observed vs. theoretical abundance and distribution of low-mass haloes) with dark-matter simulations. I asked her whether it could be a resolution issue but she said that the latest information she had is that the problem became worse with higher-resolution simulations. So, it seems to be a real problem, but on the other hand this does not prove MOND is right and DM is wrong, as Kroupa tries to spin it.

        Around the same time, there was a “great debate” (presumably to recall Shapley vs. Curtis) between Kroupa and Simon White.

        At least the cynics can no longer say that MOND research gets too little publicity in the community, suffers from lack of funding, has too few full professors working on it etc.

  4. John Hagel Says:

    Interesting twist – as I understand the above mentioned papers, the claim is not really that only MOND should replace the standard cosmological model. Rather, MOND is an example of an alternative. It may of course be the only alternative which makes sense. Still, I just checked the major research institutions in Germany and the USA as well as UK (there are not that many), and I do not actually see any professor working on MOND nor any MOND research group anywhere. Even Kroupa’s group does not seem to be dealing with MOND that much. Maybe because he can’t get funding? One of the footnotes in the Kroupa paper seems to give a hint. So Helbig’s comment concerning MOND funding is not understandable. It is easy, though, to see that there are hugely funded standard cosmological groups everywhere.

    • I think Kroupa works mainly on MOND. Maybe he keeps a low profile for PR reasons. Bob Sanders retired recently, but he worked (and still works somewhat) on MOND. Stacy McGaugh, Jacob Bekenstein, now also Gerhard Henssler and James Binney. All famous people at famous institutes working on MOND.

      • David Brown Says:

        “Hensler” is the correct spelling.

        Click to access CV.short.pdf

        I claim that the Fernández-Rañada-Milgrom effect is approximately equivalent to Milgrom’s acceleration law by an easy scaling argument. Consult any expert on general relativity theory.

  5. telescoper Says:

    I take the point about where the funding goes – cosmology has its bandwagons and there’s too little work on alternatives to the standard paradigm. Nevertheless I still think this argument is hyped.

    It would be good to get some comments from the mainstream simulation community, about this paper as I am by no means an expert. Galaxies are point masses to me.

    • Bryn Jones Says:

      Are there any detailed simulations of structure and galaxy formation under MOND? Does anybody know whether MOND is capable of forming anything like the structures seen? Does it form galaxies with anything like the mass and size distributions observed? Or would MOND fail totally at these?

      There are certainly bandwagons in science, but there are also fringe ideas that a great many scientists think are not worth spending time on.

      • John Hagel Says:

        The argument would be valid I guess if such work had been done to a level that data can be compared with. But if alternatives are not supported (one needs more than just a few dispersed professors somewhere, but larger research teams to tackle such issues) then it is strange to declare that some alternative cannot compare with the standard model since it has not been shown to agree with the data. That is, to do structure simulations in MOND, for example, one needs massive computational facilities, the computer codes, the personpower to develop these and do the computations, and the personpower to anlyse the results. But even bandwagons are usually dispersed off such that later generations even laugh at the simplistic ideas held onto by majorities despite fringe ideas proving to be the correct descriptions (e.g. famously heliocentric vs geocentric world views). So majority arguments can have no place in serious science. The above mentioned Kroupa paper does in fact discuss those cosmological computations that have been done by a few in MOND, and apparently they are a great success, even concerning the CMB and the emergence of structure at the big bang. Kroupa also lists current failures of MOND, but I am impressed how few these are in comparison to the huge number of failures of the standard model.

      • telescoper Says:

        The standard cosmology has achieved many successes, but there are some situations in which it is extremely difficult to work out the detailed consequences of the model (e.g. in galactic structure). I remain entirely unconvinced that any of the work described in these papers represents a refutation of the basic idea of cold dark matter. It does, however, indicate that we don’t understand how small-scale structures evolve in detail within the framework.

        MOND has its proponents, but I don’t think they do their argument any favours by hyping results to pretend that there’s a definitive rejection of the dark matter model.

        I still think the standard cosmology is the best model we have, and it has raised the bar in terms of the evidence needed to test it. However, that is not to say that it is proven beyond all reasonable doubt.

        I quote Alfred North Whitehead:

        There is no more common error than to assume that, because prolonged and accurate mathematical calculations have been made, the application of the result to some fact of nature is absolutely certain.

        That could certainly apply to the cosmological simulation industry!

      • stringph Says:

        You can’t do ‘cosmological calculations in MOND’ as it’s not even a relativistic theory. You can try various ad-hoc relativistic versions which have several extra fields and free functions, making them horrendously difficult to calculate with, but giving you in principle an infinite amount of freedom to fit data… Now with (relatively) simple choices for the free functions you *still* need some extra matter to fit clusters and large scale structure, for example sterile neutrinos. Any MOND-like ‘alternative’ that’s remotely viable on large scales is significantly more complicated than CDM, which has precisely one parameter.

        Two things happen much too often for me to give the dark-matter-crisis crowd much credence. One, an announcement that CDM is ruled out by some observation, but without any comparison to any actual CDM calculation including relevant uncertainties. Two, an attempt to motivate MOND-like models by highlighting a supposed problem with CDM, while omitting to mention any problems with the alternative theory.

        You can’t compare two models by applying different ad-hoc criteria to each. Pick a data set including everything that would be influenced by the difference between the models and that can be reasonably calculated. Do the calculation on the same basis for each, accounting for free parameters and uncertainties. Then compare.

        The Bonn press release is a prime example: collisions can happen just as well in CDM galaxy formation as in any other, so why not include them as a possible part of the standard model. If a collision did happen I don’t see why it should reduce our degree of belief in CDM.

        The last remark claiming to ‘rule out the presence of dark matter in the universe’ – when all they’ve studied is part of the neigbourhood of a single galaxy – is of particular fatuity.

  6. John Hagel Says:

    The last statement in the previous comment is very true. To my mind, reading through the three papers discussed here and the other papers discussed here : , the cold dark matter model is out. Note especially figure 14 in that Kroupa paper. But perhaps I am just not predisposed enough to follow the crowd.

    • telescoper Says:

      You’re entitled to your opinion, of course, but to my mind while the first of the papers you referred is interesting, the Kroupa “review” is just a silly rant. He omits most of the successes in accounting for the detailed statistics of the large-scale structure of the cosmic web and includes the known problem with small-scale structure in galaxies umpteen times as if they were independent. Pieces like that don’t advance the arguments one jot.

  7. John Hagel Says:

    Indeed, this is the sort of reaction quite typical of the cosmological community (which I am luckily not part of). Having been following the pros and cons a little over time I do note the unscientific type of debate that typically comes from the dark sector. There are various “silly” statements in recent papers I have noted (e.g. by Libeskind et al. 2011). “MOND is silly” is a favourite one. Funny I find though is that solutions, which stand up to closer scrutiny, are never offered. Instead, the arguments get entangled into mutuall inconsistencies. One of the ones pointed out by Kroupa is that the satellites of our Galaxy must have accreted at low redshift AND at high redshift simultaneously, as is shown by two different papers. It seems to me that logics or the scientific debate is not really part of the argument any longer. Kroupa appears to try to remedy this. To refer to a review, which to my mind looks quite excellent, as being “silly rant” is dismantling scientific credibility.

    • telescoper Says:

      Well, if people don’t like having their papers described as silly they shouldn’t write silly papers.

      People can read the Kroupa “review” and decide for themselves.

  8. John Hagel Says:

    I had sent an e-mail to Kroupa (easy to find) overnight and he just replied quite amused mentioning that it seems his paper is so silly that he was invited to the upcoming Marcel Grossmann meeting because of it.

    • telescoper Says:

      I understand he’s done a similar thing at conferences before. It just shows how open-minded cosmologists are. It’s good to hear the opposing arguments, even if they turn out to be not very coherent.

      It reminds me a bit of Pietronero, who went around arguing that the Universe was a fractal some years ago. He got invited to big meetings too, even though nobody believed him.

      • John Hagel Says:

        Well, you are displaying a strange sort of logics. It is quite possible that arguments in the end turn out to be invalid, and time will tell concerning standard cosmology (although for me the situation is already clear despite the large scale data, which by the way are mentioned by Kroupa). But, invitations to such major conferences are not based on “silly rants”. At least not in my experience. Why waste the time of the important colleagues with a quack who rants sililiy through space and time? It would be a waste of public money. In any case, mentioning “believe” is again pointing to a disquitening issue with you: _modern_ science is not about believe, but about evidence. If you cannot separate this, then the question becomes who the real silly ranter is? Pietronero may have had a hypothesis, which was interesting and so he was asked to give presentations, and in the end the data may have invalidated the hypothesis. To then think badly of him _just because of this_ is positively unscientific. This alone would not make him silly. Or are you suggesting that the scientific progress may only be based on “real scientists” (however you would define them but I guess you see yourself in this group) following their believes? We had this for centuries in the past…

      • telescoper Says:

        Reasoned belief based on evidence is exactly how science works. I’m surprised you think otherwise.

        The point I was making was that the cosmological community quite often arranges these debates, even when a minority view is being represented. I don’t think badly of Pietronero for taking a contrary view. I do object to the way he mishandled the evidence to fit a preconceived idea.

        Now I’d be pleased if you would reply to the email I sent to the address you give while posting comments.

    • telescoper Says:

      PS. I imagine you did find it very easy to contact him, given your location. I’m even starting to wonder whether you’re a member of his group!

      • John Hagel Says:

        I wish I were, as that group seems to be doing really interesting research (mostly not on cosmology though, as I can see from his web page). As you will know a location you see does not imply a physical location of the correspondent. But, I do recommend to you that you stick to scientific arguments rather than diverting from the real issue at hand by the non-scientific statements that you have now been making too many times.

      • telescoper Says:

        This is a personal blog. I’ll say whatever I want thank you very much.

        I also have rules for posting comments – I’d be grateful if you would comply by replying to the email I sent you.

  9. telescoper Says:

    I remind commenters that I do not accept pseudonymous or anonymous comments and require a valid email address for confirming identity. Failure to comply with these simple rules will result in your comments being blocked.

    Comments from colleagues and associates of the authors of the paper discussed in this post will only be accepted if posted under their real names.

  10. Yes, it is true that many more CDM simulations have been done than MOND simulations. Although the metaphor isn’t appropriate here, maybe someone who has a penchant for dark horses could fund a huge MOND effort and see what comes of it.

    On the other hand, one can’t complain that a fair comparison is impossible because of lack of resources for MOND and at the same time claim that CDM is ruled out.

    Maybe the successes of MOND will go away once detailed computations are made. 🙂

    • stringph Says:

      The problem is that there are many different self-consistent models which produce something like ‘MOND’ – whereas MOND itself is not self-consistent, by most criteria for a decent dynamical theory of cosmology – Lorenz covariance, equivalence principle… MOND-like models are generically complicated and even contain free fitting functions, so (even if you can deal with the complexity) it’s not a case of setting up the simulation once and for all. Some people are making brave efforts though.

  11. Dave Carter Says:

    Kroupa may have overstated his case, but I would defend him for doing so, given the number of times you see statements in papers like “our observations support the CDM/Hierarchical model of galaxy formation”. Sorry, which other models of galaxy formation have you addressed and in what way do your observations argue against them? Possibly there are even such meaningless statements of support for CDM in papers I am on.

    I must admit that I find the faith in CDM, perhaps more particularly the hierarchical models that come out of the CDM hypothesis, a little misplaced. Not saying that I have the answers, but there are a whole host of properties of galaxies on a smaller scale which do not come naturally out of the models. Things like the numbers of low-mass galaxies, the strength of colour gradients, the specific frequency of globular clusters, the UV excess in massive ellipticals, the prevalence of nuclear clusters in low-mass galaxies. All of these, individually, can be fudged to fit. But the need for the fudges points to a more fundamental problem, in my own opinion.

    As far as gravity is concerned, someone else who is working on alternative models, and who is rarely cited by the mainstream, is Philip Mannheim at Connecticut.

    • telescoper Says:

      Indeed, there is a problem with CDM on small scales. But whether the problem is with CDM itself, or with our ability to work out what it predicts, is by no means clear.

  12. stringph Says:

    By the way, there is a newish review article out by some rather less polemic proponents of non-CDM dynamics: see arXiv:1112.3960.

    One nice paper from some time ago using SDSS satellite galaxy dynamics to challenge MOND was this one :

    – its citations are interesting too (check the ‘cited by’ link on that page).

  13. Bloody Orphan Says:

    As far as I can see, both MOND and DM are not mutually exclusive.

    MOND is simply describing the variablility of DM in our Universe.

    The Higgs, and recent fluctuations in the Gravity Constant experiment both point to the validity of MOND calculations, and at some point in the future it will integrate seemessly with our current Standard Model, including Dark Matter interactive coefficients.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: