The Most Interesting Galaxies Are SMUDGES

Ultra Diffuse Galaxies remain a very hot topic in astronomy. You know the drill by now : great big fluffy things with hardly any stars and sometimes little or no dark matter, not really predicted in numerical simulations. I'm not going to recap them again because I've done this too many times, so I leave it as an exercise for the reader to search this blog and learn all about them. Get off yer lazy arses, people !

UDGs were first found in clusters but have since been found absolutely everywhere. Why clusters ? Well, because they're so faint, getting redshift (i.e. distance) measurements of them is extremely difficult. This means their exact numbers are fiendishly difficult to characterise : without distance you can't get size, which is one of their distinguishing properties – so without size you can't even count them. And if you can't count them, you can't really say much about them at all.

Getting distances in clusters, however, is much easier. There the distance to the whole structure is anyway known. The first studies found lots of UDG candidates in clusters but very few in control fields, so most of those are certainly cluster members rather than just being coincidentally aligned on the sky. Of course it's always possible that a small fraction (at the few percent level or less) weren't really in the cluster and therefore not truly UDGs, but statistically, the results were definitely reliable.

The SMUDGES project (Systematically Measuring UDGs) is a major effort to begin to overcome the limitation of relying on clusters for distance estimates*. In essence, they try to develop a similar procedure for clusters but which can be applied to all different environments. They want results which are at least statistically "good enough" to estimate the distance, even if there's some considerable margin of error. 

* The main alternative thus far has been gas measurements, which give you redshift without relying on the much fainter optical data. This, however, has its own issues.

This paper is mainly a catalogue, and to be honest I rarely bother reading catalogue papers. In fact I only read this one to see what low-level methods they used to do the size estimates, since we have some possible candidate UDGs of our own we want to check. But as it turned out, they also present some interesting science as well, so here it is.

Most of the paper is given to describing these methodologies and techniques. It's pretty dry but important stuff, and like with the first cluster-based studies, they can't be sure that absolutely every candidate they find is really a UDG. Actually these measurements are, inevitably, quite a lot less reliable than the cluster studies, but they're careful to state this and the results are still plenty good enough to identify interesting objects for further study.

One interesting selection effect they note early on is that studies of individual objects tend to overestimate their masses (compared to studies of whole populations), since these tend to be particularly big, bright, and prominent. This at least helps begin to explain why some division has arisen in the community regarding the nature of UDGs : the objects studied by different groups are similar only at a broad-brush level, and in detail they may have significant differences. That's not an explanatory bias that was obvious to me, but maybe it should have been. It seems perfectly sensible with hindsight, at any rate.

And, once again, this is another study where the authors resort to flagging dodgy objects by eye, in another example of how important it is to actually look at the data. The machines haven't replaced us yet.

I won't do a blow-by-blow description of their procedures this time, but their final catalogue comprises about 7,000 objects, which they supplement with spectroscopic data where available. One of the main topics they address is the big one : what exactly are UDGs ? Are they galaxies with normal, massive dark matter halos but few stars, or do they instead have weird dark matter distributions ?

They conclude... probably the former. But this is not to say that they are "failed Milky Way" galaxies that have just not formed many stars for some reason : at the upper end they're probably still a few times less massive than that, and at the lower end that might be more than a factor ten difference. So mostly dwarf galaxies, but with normal dark matter distributions and very few stars. They get mass estimates from a combination of counting the number of globular clusters, which correlates with the total halo mass in normal galaxies, and their own statistical method to estimate other galaxy properties (which I don't fully understand). 

These relations don't always work well, however, sometimes experiencing "catastrophic failure", by which they mean errors of an order of magnitude or more. Why this should be is impossible to say at this stage, but, intriguingly, might point to the dark matter distribution being indeed different in UDGs compared to normal galaxies, at least some of the time. Overall though this appears unlikely, because to make this work with the observed scaling relations, the dark matter would have to be more concentrated than expected, even though the stars are the exact opposite : much more spread out than usual.

Bottom line : they think UDGs are mainly dwarf galaxies (though a few may be giants), with normal dark matter contents but very poor star formation efficiency for whatever reason. I'm not so sure. They say the distribution of some parameters (e.g. stellar mass within a given radius) is the same for both UDGs and other galaxies but to me they look completely different; it doesn't help that the figure caption states two colours when there are clearly three actually used. What's going on here I don't know, but very possibly I've missed something crucial.

Of course this paper won't solve anything by itself, but it gives a good solid start for further investigations. As with the previous post, this is another example of how important it is to classify things in a homogenous way. At least one SMUDGES object is found within our own AGES survey fields, and was in fact known to much earlier studies. Sometimes what can look at first glance to be a normal object actually turns out to be something much more unusual, but it's only when you have good, solid criteria for classification that this becomes apparent. 

Which is all very good news for AGES. I suspect there are actually quite a lot more UDGs lurking in our data. All we need is a team of well-armed and angry postdocs to track them down... i.e. a great big healthy grant. Well, a man can dream.

Dey's Blue Blobs

Today's paper is more exciting than I can fully let on.

In the last few years there have been a handful of seemingly-innocuous discoveries in Virgo that don't quite fit the general trends for normal galaxies. They're very faint, very blue, metal-rich*, and some are incredibly gas-rich. The most convincing explanation thus far is that they're ram pressure dwarfs : not galaxies exactly, but bound systems of stars that formed from condensations of gas stripped by ram pressure. 

* Meaning they have lots of chemicals besides hydrogen, because astronomers have weird conventions like that.

The advantages of this explanation is that ram pressure is a high speed phenomenon, so could easily explain why the objects are so far from any candidate parent galaxies (tidal encounters can do this too, but usually require lower interaction velocities), as well as why they're so metal-rich. Primordial gas is basically nothing but hydrogen and helium, and to get complex chemistry you need multiple cycles of star formation, which makes it virtually certain that the gas here must have originated in galaxies. Why exactly they've only just started forming stars is unclear, though it's possible they do have older stellar populations which are just too faint to identify. And these things really are faint, with just a few thousand solar masses of stars... in comparison to the usual millions or billions expected in normal galaxies.

One of the main problems in understanding these objects has been the understandably crappy statistics. With only a half-dozen or so objects to work with, any conclusions about the objects as a population are necessarily suspect. That's where this paper comes in.

Finding such objects isn't at all easy. They're difficult to parameterise and tricky for algorithms to handle, so they opt for a visual search. And quite right too ! Humans are very, very good at this, as per my own work (which I'll get round to blogging soon). Having just one person run the search would risk biases and incompleteness, so they use a citizen science approach based on Galaxy Zoo. 

The result was a total of nearly 14,000 "blue blob"* candidates. But this is being extremely liberal, and many of these might just be fluff : noise or distant background objects or whatever. A more rigorous restriction in which at least three people had to identify each candidate independently reduces this to just 658. Further inspection by experts trimmed this to 34 objects – a still more than respectable improvement over previous studies. And while I previously berated them for claiming that the objects only exist in clusters without having looked elsewhere, this time they at least looked at Fornax as well as Virgo. Fornax is another cluster, but interestingly no candidates were found there.

* C'mon guys, this is the name we're going with ? Really ? Oh. Well, fine. Suit yourselves.

But they don't stop with the results of the search. They cross-correlate their results with HI gas measurements from ALFALFA and, yes, AGES (thanks for the citations, kindly people !), and also observe eight of them with the 10m-class Hobby-Eberly Telescope for spectroscopy of the ionised gas. This is extremely useful as it provides a robust way of verifying that these objects are indeed in the cluster and not just coincidentally aligned, and also shows the the gas in the objects is being affected by the star formation.

Let me cover the main conclusions before I get to why I'm so excited by this work. First, their findings are fully consistent with and support the idea that these are ram pressure features. Their spectroscopy confirms the high metallicity of the objects, comparable to tidal dwarfs – so they have indeed formed by material which was previously in galaxies. They avoid the very centre of the cluster (where they'd likely be rapidly destroyed) and are preferentially found where ram pressure is expected to be effective. 

There's also an interesting subdivision within these 34 candidates. 13 of these are "rank 1", meaning they are almost certainly Virgo cluster objects, whereas the others are "rank 2" and are likely to have some contamination by background galaxies. Most of the rank 2 objects follow the general trends in colour and magnitude as for normal galaxies, but the rank 1 are noticeably bluer. They're also forming stars at a higher than expected rate (though, interestingly, not if you account for their total stellar mass). So indeed these are galaxy-like but not at all typical of other galaxies : they are galaxian, not galaxies.

Now the fun stuff. They identify two supposedly optically dark clouds I found in Virgo way back when and have since based most of my career on, hence – exciting ! They do have optical counterparts after all, then. Actually, one these is relatively bright, and I suggested it as a possible counterpart back in 2016. But it wasn't convincing, and its dynamics didn't seem to match well at all. These days of course everyone is all about the weird dynamics, but back then this seemed like a pretty good reason to rule it out. Since then, our VLA data has independently confirmed the association of the stars and the gas, and Robert Minchin is writing that one up as a publication.

That object has about twenty times as much gas as stars. The second object is altogether fainter, having a thousand times or more gas than stars ! Even with our VLA data we couldn't spot this*, and I probably wouldn't even believe this claim if they didn't have the optical spectroscopy to support it. It looks likely that in this case we're witnessing the last gasp of star formation, right at the moment the gas dissolves completely into the cluster.

* The VLA data has much better resolution than the original Arecibo data, so it can localise the gas with much greater accuracy and precision. This means that it can show exactly where the HI is really located, so if there's even a really pathetic optical counterpart there, we can be confident of identifying it. But of course, that counterpart must be at least visible in the optical data to begin with.

While they comment directly on two of our objects, they actually implicitly include another three measurements in the table. We never identified these as being especially weird; they just look like faint blue galaxies but nothing terribly strange. And that really underscores the importance of having enough resources to dedicate to analysing areas in detail, which, frankly, we don't. It also shows how important it is to quantify things : visual examination is great for finding stuff, but it can't tell you if an object is a weird outlier from a specific trend. Even more excitingly, almost certainly it means that there are a lot more interesting objects in our data that have already been found but not yet recognised as important.

But the most fun part came from doing a routine check. Whenever anyone publishes anything about weird objects in our survey fields, I have a quick look to see if they're in our data and we missed them, just in case. Every once in a while something turns up. This is very rare, but the checks are easy so it's worth doing. And this time... one of the other blue blobs has an HI detection in our data we previously missed.

Which is very cool. The detection is convincing, but there are very good reasons why we initially missed it. But I don't want to say anything more about it yet, because this could well become a publication for my PhD student. Watch this space.

Taking galaxies off life support

Very long-term readers may remember my anguished efforts (almost a decade ago) to build a stable disc galaxy. Sweet summer child that I was, I began by trying to set up the simulations to just have gas or stars, but no dark matter. I thought – understandably enough – that adding more components would just make things more complicated, so best to start simple. I was planning to gradually ramp up the complexity so I could get a feel for how simulations worked, eventually ending up with a realistic galaxy that would sit there quietly rotating and not hurting anyone.

That wasn't what I got. Instead of a nice happy galaxy I got a series of exploding rings instead. Had that been a real galaxy, millions of civilisations would have been flung off into the void.

It turns out that dark matter really is frightfully necessary when it comes to keeping galaxies stable. Dark matter is a galaxy's emotional support particle, preventing it from literally flying apart whenever it has a mild gravitational crisis. Stable discs are easy when you have enough dark mass to hold them together.

(Of course, this is only true in standard Newtonian gravity. Muck about with this and you can make things work without dark any matter at all, but I'm not going there today.)

You don't always need dark matter to keep things together though. Plenty of systems manage just fine without it, like planetary systems and star clusters. But it's come as a big surprise to find that there are in fact quite large numbers of galaxies which have little or no dark matter, a result which is now reasonably (and I stress that this is an ongoing controversy) confirmed. We always knew there'd be a few such oddballs, if only from galaxies formed from the debris of other galaxies as they interact. But nobody thought there'd be large numbers of them existing in isolation. So what's going on ?

Enter today's paper. This is one in a short series which to be quite honest I'd completely forgotten about, partially because the authors forgot to give the galaxy a catchy nickname. Seriously, they could learn a lot from those guys who decided to name their galaxy Hedgehog for no particular reason. I'm only half-joking here : memorable names matter !

But anyway, this was an example of a UDG with lots of gas that appeared to have no dark matter at all. I wasn't fully convinced by their estimated inclination angle though, for which even a small error can change the estimated rotation speed and thus the inferred dark matter content substantially. A independent follow-up paper by another team ran numerical simulations and found that such an object would quickly tear itself to bits, whereas if if was just a regular galaxy with a very modest inclination angle error then everything would be fine. And there have been many other such studies of different individual objects, all of them mired in similar controversies. 

Since then, however, I've become much more keen on the idea that actually, a lot of these UDGs really do have a deficit or even total lack of dark matter after all. The main reason being this paper, which is highly under-cited in my view. Now it's entirely plausible that any one object might have its inclination angle measured inaccurately*. But they showed that the inclination-corrected rotation velocity of the population as a whole shows no evidence of any bias in inclination. Low inclinations, high inclinations, all can give fast or slow rotating galaxies, consistent with random errors. That some show a very significant lower than expected rotation therefore seems very much more likely to a a real effect and not the result of any systematic bias.

*Though all of these terms like "bias", "errors" and "inaccuracies" are, by the way, somewhat misleading. It's not that the authors did a bad job, it's that the data itself does not permit greater precision. That is, it allows for a range of inclination angles, some of which lead to more interesting results than others. The actual measurements performed are perfectly fine.

What about that original galaxy though ? AGC 11405 might itself still have had a measurement problem. Here the original authors return to redress the balance.

It seems that in the interim I missed one of their other observational papers which changes the estimates of exactly how much dark matter the galaxy should have; probably this is lost somewhere in my extensive reading list. The earlier simulation paper found that the object could be stable only (if at all) with a rather contrived, carefully fine-tuned configuration of dark matter, and there wasn't any reason to expect such a halo to form naturally. Couple that with the findings that it could easily be a normal galaxy if the inclination angle was just a bit off, and that made the idea of this particular object seem implausible, even if a population of other such objects did exist.

But that interim paper changes things. Whereas previously they used the gas of the object to estimate the inclination angle, now they got sufficiently sensitive optical data to measure it from the stars, and that confirms their original finding independently. They also improved their measurements of the kinematics from the gas, finding that it's rotating a bit more quickly than their original estimates, meaning it has a little bit more scope for dark matter. More significantly, the same correction found that the random motions are considerably higher than they first estimated.

What this means is that the dark matter halo can be a bit more massive than they first thought, and the disc of the galaxy doesn't have to be so thin. A thick disc with more random motions isn't so hard to keep stable because it's fine if things wander around a bit. So they do their own simulations to account for this, with the bulk of the paper given to describing (in considerable detail) the technicalities of how this was done.

They find that an object with these new parameters can indeed be stable. Rather satisfyingly, they also run simulations using the earlier parameters, as the other team already did independently. And they confirm that with that setup, the galaxy wouldn't be stable at all. So the modelling is likely sound, it's just that it depends quite strongly on the exact parameters of the galaxy. They confirm this still further with analytic formulae for estimating stability, showing that the new measurements of the rotation and dispersion are, once again, predicted to be stable.

But if the galaxy actually does have a hefty dark matter halo after all, doesn't that mean it's just like every other galaxy and therefore not interesting ? No. As far as I can tell, the amount of dark matter is still significantly less than expected, but also its concentration (essentially its density) is far lower : a 10 sigma outlier ! So yes, it's still really, really weird, with the implied distribution of dark matter still apparently very contrived and unnatural.

So how could such a galaxy form ? That's the fun part. It's important to remember that just because dark matter doesn't interact with normal matter except through gravity, this is not at all the same as saying it doesn't interact at all ! So some processes you'd think couldn't possible affect dark matter... probably can*. Like star formation, for instance. Young, massive stars tend to have strong winds and also like to explode, which can move huge amounts of gas around very rapidly. It's been suggested, quite plausibly, that this is what's responsible for destroying the central dark matter spikes which are predicted in simulations but don't seem to be the case in reality. The mass of the gas being removed wouldn't necessarily be enough to drag much dark matter along with it, but it could give it a sufficient yank to disrupt the central spike.

* And it's also worth remembering that just because dark matter dominates overall, this isn't at all true locally. This means that movement of the normal baryonic matter can't always be neglected. 

The problem for this explanation here is that the star formation density must be extremely low to get objects this faint. So whether there were ever enough explosively windy stars to have a significant effect isn't clear. Quantifying this would be difficult, especially because dwarf galaxies are much more dominated by their dark matter than normal galaxies – yes, they'd be more susceptible to the effects of massive stars because they're less massive overall, but the effect on the dark matter might not necessarily be so pronounced.

The authors here favour a more exotic and exciting interpretation : self interacting dark matter. The most common suggestion is self-annihilating dark matter that's its own anti-particle, which would naturally lead to those density spikes disappearing. There could be other forms of interaction that might also "thermalize" the spike... but of course, this is very speculative. It's an intriguing and important bit of speculation, to be sure : that we can use galaxies to infer knowledge of the properties of dark matter beyond its mere existence is a tantalising prospect ! But to properly answer this would take many more studies. It could well be correct, but I think right now we just don't have enough details of star formation to rule anything out. Continuing to establish the existence of this whole unuspected population of dark matter-deficient galaxies is enough, for now, to be its own reward.