Apologies to my regular readers for disappearing last week, but I was away and then super busy with an important deadline. Good news is that I have managed to put together my top arXiv column for both week 6 and week 7, so you get 2 at the price of 1. Enjoy!
#1 2002.00459: Annual modulations from secular variations: relaxing DAMA? by Dario Buttazzo et al.
The question of what the hell the DAMA experiment has been seeing for about 20 years now is probably one of the biggest mysteries in the field of dark matter (DM), though not necessarily one people care about anymore (more later). If DM in the form of WIMPs is present in the galactic halo, the motion of the Earth around the Sun should generate an annual modulation in the event rate seen in direct detection experiments, peaking at June 2 in the Northern hemisphere. This was in fact first worked out in this seminal paper by my PhD advisor Katie Freese together with Andrzej Drukier and David Spergel in the 80s. The DAMA experiment was set up to look for this annual modulation, and for about 20 years they have been seeing such a signal which appears at first glance perfectly compatible with DM (the phase is spot on), at a number of sigmas so high (>10) that it is even pointless to quote it. Trouble is that pretty much any experiment which came after DAMA has ruled out their claimed detection at equally many sigmas, and whereas crazy DM (with colourful-sounding names such as Xenophobic DM - no kidding!) or background models for reconciling everything together have been put forward, none appears to be plausible. It doesn’t help that the DAMA collaboration has over and over refused to make their data public. Because of this, the attitude of the community has slowly been moving towards ignoring DAMA (which explains why above I said that most people don’t care anymore). However, DAMA remains in some sense the elephant in the room. Could it be that the solution to this mystery is simply a poor analysis method?
This week’s paper looks at precisely this possibility, i.e. that the way DAMA computed/presented their modulation results might be dangerous and introduce a spurious modulation signal when really there wasn’t any. To give some context, what does DAMA do? The modulating residuals they show are computed by yearly subtracting (starting from about the same date) the weighted average of the total rate over the year. This is dangerous, because you by doing so you are introducing a somewhat preferred time-scale in the system (it’s a sort of Nyquist frequency in some sense) which exactly corresponds to the time-scale of the DM signal you want to detect. If the underlying unmodulated rate were time-varying, this time variation could spuriously be translated into an annual modulation due to this dangerous procedure. In practice you can transform a linearly growing rate into a sawtooth which under certain conditions can appear like the cosine signal DAMA claims to see. It is worth reading section 2.1 of the paper, where the authors present an illuminating example showing the danger of such a procedure in a bibliometric setting. In this sense, the safer way of analyzing the data would be to model the time dependence of the background across the whole lifetime of the experiment, or at the very least do this dangerous background subtraction over a different time period. The authors then study a number of plausible time-variations in the unmodulated rate which could potentially reconcile DAMA with the rest of the world, finding that a slowly steadily growing background could potentially reproduce DAMA’s signal. However, it is unclear whether this growth could come from some plausible background, and the situation is overall inconclusive. The main point of the paper, nonetheless, is a very valuable one: it could be that a spurious signal could be coming from a poor background treatment of a type not considered so far (people had focused, for obvious reasons, on annually modulating backgrounds). And the nice thing is that, if the “mistake” proposed by the authors were the reason behind DAMA’s claim, it would be very easy for the collaboration to figure this out simply by looking at the complete time-dependence of their full (modulated+unmodulated) rate. But of course, such a full rate has not been made public, as with all things DAMA…(although in principle the collaboration could easily check this even without making the data public!)
#2 2001.11784: Emergence of ghosts in Horndeski theory by Eugeny Babichev
I already wrote about Horndeski gravity in my week 3 post, recall it is the most general 4D scalar-tensor theory with second-order equations of motion (sufficient condition for avoiding the Ostrogradsky ghost). The common lore in the field of gravitation is therefore that Horndeski gravity is a “healthy” theory, and it is not surprising that a huge chunk of the community is working on it. The question is then whether, even though the theory is apparently healthy, it can still become pathological “on shell”, i.e. along a specific solution to the equations of motion, even if it starts healthy? This is the question Babichev sets out to answer in this week’s paper.
Specifically, Babichev considers a concrete subset of Horndeski gravity consisting of a K-essence term (i.e. a non-linear functional of the scalar field kinetic term) and a cubic galileon term (i.e. the kinetic term multiplied by the Laplacian of the scalar field). The theory is known to give rise to caustic singularities, potentially dangerous. However, even before the appearance of the caustic one can study the perturbations of the theory at second order, construct the Hamiltonian, and find that at some point along the solution to the equations of motion it becomes unbounded by below, usually a sign of ghost instabilities. In fact, Babichev explicitly shows that changing the sign of the initial condition for the velocity of the field can definitely lead to the appearance of a ghost instability. Does this mean that the theory is unhealthy, the specific solution is unhealthy, or the specific initial conditions are unhealthy? I would go for the latter. The result is really interesting and the idea partially resonates with the same idea I blogged about in the first week post regarding the finite action principle, i.e. that one should really look at the behaviour of solutions to the equations of motion to assess the healthiness of a theory rather than just the equations of motion themselves or even just stop at the action. One point to note is that in his study Babichev assumes a non-dynamical metric, i.e. the latter is not a variable of the theory. I do not have a feel for whether the appearance of the ghost instability depends crucially on this somewhat unrealistic assumption, but it would be definitely worth relaxing such an assumption in the future.
#3 2002.02102: Self-Interacting Dark Matter and the Origin of NGC1052-DF2 and -DF4 by Daneng Yang, Hai-Bo Yu, and Haipeng An
NGC1052-DF2 is an ultra-diffuse galaxy which is part of the NGC1052 group, brought to fame by Pieter Van Dokkum’s group in a paper claiming this galaxy is devoid of dark matter (DM). Shortly afterwards, NGC1052-DF4 was found to have similar characteristics. Is the situation clear and the community undivided on this result? No. Does this spell the end of particle DM and the victory of models such as MOND? Probably not, because they could have lost most of their DM from tidal stripping in the host’s tides. If correct, can it teach us something about the interactions of DM? Most likely! It is this third route that Yang et al. follow in this week’s paper. What they basically try to show is whether these two galaxies can teach us something about cold DM (CDM) versus self-interacting DM (SIDM), and the results appear to be very promising. What is the fundamental difference between CDM and SIDM in this context? The halo profile. If you start with a cuspy NFW-like profile, it will remain such in CDM, whereas with SIDM the inner halo can thermalize, which leads in a quite natural way to a cored profile (which is why SIDM is often advocated as a solution to the core-cusp problem). The authors therefore study a NGC1052-like system, including its satellite galaxies, to see whether and SIDM can reproduce the observed DM-less galaxies.
The final result is yes, although CDM halos require somewhat unrealistically low halo concentrations (halo concentration is a measure of, no surprise, how concentrated the halo is, as it essentially measures the radius at which the halo contains a given amount of mass in its interior). The reason is that the cuspy CDM halos are more resistant to tidal stripping, meaning that you have to start with a less concentrated halo to begin with if you want to get the same result as SIDM, which instead helps tidal stripping by leading to a softer halo profile which pushes DM towards the outer regions. The authors also tried a third model, i.e. CDM+baryonic feedback, which has also been advocated to solve the many small-scale problems, but they find that in order to match the observed properties of the two DM-less galaxies requires a stellar distribution which does not match that predicted by CDM models with strong baryonic feedback. This paper’s results are very interesting although I would say they are in some way quite preliminary. There are a few assumptions which probably should be revisited in future work (e.g. all satellites are confined to radial orbits in a plane, which certainly help tidal stripping). Naïvely one might think that the DM self-interaction cross-sections taken would appear to be excluded by the Bullet cluster: the authors take 3 cm2/g and 5 cm2/g, whereas the Bullet Cluster appears to exclude anything above 0.1 cm2/g. However, in SIDM models with light mediators the cross-section is very naturally velocity-dependent and tends to decrease towards cluster scales (here we are looking at galactic scales), plus there are indications that a more conservative analysis of the Bullet Cluster might lead to a limit closer to 2 cm2/g. Technical details aside, the idea behind this paper is very interesting, and I would definitely like to see more follow-up ideas on how DM-less galaxies might teach us something about the particle properties of DM, much as that might sound an oxymoron!