This week’s entry discusses two papers on shadows (one in relation to superradiance, and one in relation to naked singularities), and one paper addressing the important question of how model-independent BAO measurements are. Following the discussion of the third paper, I have included a long-ish bonus discussion on what my opinion is on this whole BAO model-independence business (some readers might find the discussion provocative, if so that was not my intention). Enjoy!
#1 2004.05178: Evolution of black hole shadows from superradiance by Gastón Creci, Stefan Vandoren, and Helvi Witek
The mechanism of superradiance, which in the specific context of black holes (BHs) is often referred to as “black hole bomb“, is an effect by which a bosonic field scattering off a rotating black hole can be amplified at the expense of the BH’s mass and angular momentum. The amplification can then be repeated and lead to a run-away growth of the bosonic field, with explosive consequences. Superradiance therefore opens the possibility of using BHs to study ultra-light bosonic fields, such as axion-like particles which could comprise the dark matter of the Universe. Thanks to the spectacular image delivered by the Event Horizon Telescope last year, we now have, besides gravitational waves, a new window onto the physics of BHs, namely BH shadows. Plagiarizing what I wrote in my Week 15 entry, a BH shadow is the apparent (i.e. gravitationally lensed) image of the photon sphere (the region of space-time where gravity is so strong that photons travel in unstable, not necessarily circular, orbits). In other words, the BH shadow is the closed curve on the sky separating capture from scattering orbits. Some earlier works, more precisely 1509.00021 and 1909.08039, had focused on studying the effect of superradiance on BH shadows, but only considering the final state of superradiant instability, i.e. a slowly dissipating bosonic cloud. Another recent work, 1906.03190, studied instead the evolution of the size of the BH shadow in the presence of superradiant instability.
In this week’s paper, Creci and collaborators work along the lines of the latter earlier study, combining a computation of superradiant evolution in the quasi-adiabatic approximation with the computation of shadows of Kerr BHs within a wide range of parameter space. Overall, they confirm the general result of 1906.03190 identifying two competing effects: the decrease of the BH mass would tend to make the shadow shrink (for a Schwarzschild BH the size of the shadow is 3*sqrt(3) M, with M the BH mass, and for a Kerr BH the shadow is only slightly smaller and importantly more oblate). On the other hand, the decrease in BH angular momentum tends to make the shadow larger. Which of these two effects wins depends partly (but not completely) on the size of the so-called gravitational coupling alpha, the ratio between the BH gravitational radius and the Compton wavelength of the bosonic field. However, the overall changes in the shadow size are of order μas, i.e. just below the current resolution of the Event Horizon Telescope, and most importantly occur on timescales which surpass by several orders of magnitude any conceivable human timescale (we are talking of about 10000-100000 years). While unfortunately this work has confirmed that, at least with current tools, being able to appreciate changes in the size of BH shadows due to superradiant instability appears unfeasible, it has also shown that superradiance leads to a host of rich effects on the evolution of BH shadows: hopefully, within these, there might possibly be something within the reach of near-future VLBI experiments?
#2 2004.06525: Shadow of a Naked Singularity without Photon Sphere by Ashok B. Joshi et al.
Earlier, when discussing paper #1, I wasn’t totally honest. In particular, when telling you that a BH shadow is the apparent image of the photon sphere (PS). Well, at least I was more honest than your average popular media outlet, many of which very incorrectly try and convince you that the BH shadow is showing you the event horizon (please don’t fall for that!). In fact, in principle objects without event horizons can still cast a shadow, as long as their space-time possesses a PS, i.e. a region of space-time where gravity is so strong that photons travel in orbits (circular in the case of the Schwarzschild BH). These two images are, in my opinion, an excellent representation of where the BH shadow comes from, for both a Schwarzschild (non-rotating) and Kerr (rotating) BH: Schwarzschild image and Kerr image (for the record, the post where these images comes from is definitely worth reading if you want to learn more about BH shadows); you can also have a look at pages 1-3 of these slides of mine. Anyway, so much to say that the common lore is that to get a shadow all you need is a PS. There have been other recent studies (e.g. 2003.06810) showing that in principle even in the absence of a photon sphere, a thin shell of matter can cause a shadow. In this week’s paper, Joshi and collaborators go one step further and show that a particular naked singularity spacetime with no PS nor thin shell of matter, can still produce a shadow.
First of all Joshi et al. derive a new naked singularity solution to Einstein’s equations, which resembles the Schwarzschild metric at large distances. They then show that this space-time can indeed produce a shadow. Explaining how this comes about in non-technical terms accessible to readers of this blog is not feasible, but a picture is worth a thousand words, so the interested reader should have a look at Fig. 1f of this week’s paper (and compare that with those two figures I linked earlier), which shows how the naked singularity actually manages to cast a shadow. Importantly, at a given mass M, the shadow size is about a factor of 3*sqrt(3) smaller than that which a Schwarzschild BH with the same mass casts (see e.g. Figs. 2b vs 2f). This provides in principle a straightforward way to distinguish between a BH and a naked singularity provided one can determine the mass of the object in question to high fidelity. Unfortunately, current BH mass determinations suffer from >100% systematics: if you want to find out why, you can read Section 3A of my recent paper with Cosimo Bambi and Luca Visinelli 2001.02986. An interesting possible follow-up study might try to include quantum gravity effects, which should become important near high-curvature singularities (possibly to cure these singularities themselves), and might manifest themselves in the shadow. Still with Cosimo, Luca, and my PhD advisor Katie Freese we looked at something like this in 1904.12983, with a very crude parametrization of quantum gravity effects curing the otherwise naked singularity caused by an object spinning above the Kerr limit (so-called “superspinars”), see also 0812.1328. It might be interesting to try and do something similar within this new naked singularity space-time found by Joshi and collaborators.
#3 2004.07263: Robustness of baryon acoustic oscillations constraints to beyond-ΛCDM cosmologies by José Luis Bernal et al.
Baryon acoustic oscillations (BAOs) are the imprint of sound waves set up in the early Universe by the interplay between radiation pressure and gravity. They show up both in the CMB power spectrum (as the damped acoustic peaks you probably have seen), and in the late-time clustering of the large-scale structure (LSS), as a slightly enhanced clustering in tracers of the LSS (such as galaxies) at a particular scale. By exploiting this excess clustering, which can be identified cleanly by computing the 2-point correlation function of LSS tracers, BAOs have been used as statistical standard rulers to measure the expansion history of the Universe at late times. In particular, if one can clearly separate distances which are transverse or along the line-of-sight (something which became possible towards the final data releases of the BOSS survey), then one can measure the angular diameter distance and expansion rate at a given redshift (which is the effective redshift of the galaxy sample one used to extract the BAO signal). In this way, BAO have been instrumental in establishing the concordance ΛCDM model. In recent years, the role of BAO measurements has been very important, particularly in relation to the H0 tension. As I wrote in many earlier posts, BAO measurements (in conjunction with SNe data) can be used to construct a so-called inverse distance ladder, which leaves little room for deviations from ΛCDM in the late-time expansion rate H(z), leading many to conclude that a solution to the H0 tension probably occurs pre-recombination. This conclusion is obviously true insofar as BAO measurements can be considered model-independent and robustly used to test (read: constrain) beyond-ΛCDM models. People in the field have mostly tacitly assumed that this is the case, and while both CMB and local distance ladder measurements have been pointed the finger at in the search for a solution to the H0 tension, BAO have escaped any sort of blame. Is this actually well justified?
In this week’s paper, Bernal and collaborators try to partially answer this question. The trouble when it comes to BAO is that there are many assumptions one has to make at all stages of a standard BAO analysis, and at some stage one inevitably assumes a fiducial set of cosmological parameters and, more importantly, a fiducial model (which is always ΛCDM). The standard BAO analysis relies on fitting, using rescaling parameters, templates constructed from ΛCDM. In practice doing so is making two important assumptions, which were recently very well summarized on top of Page 2 of 1906.03035 by Carter et al.: 1) differences in the distance–redshift relationship between true and fiducial cosmology are assumed to match a linear scaling, and 2) differences in comoving clustering between true and fiducial cosmology are assumed to be removed by the same set of free parameters used to null the non-BAO signal in the correlation function or power spectrum. One way of phrasing potential concerns related to violating 1) or 2) is: are the templates and rescaling parameters we use flexible enough to account for possible deviations beyond ΛCDM? Bernal et al. focus on three models which modify the early expansion rate in an attempt to address the H0 tension: extra relativistic degrees of freedom, early dark energy, and dark neutrino interactions. In doing so, they are essentially mostly addressing point 2) I mentioned above (whereas point 1) is mostly important for models which modify the late-time expansion). They find that the bias due to residual model-dependence is at most 1 sigma for the dark neutrino interactions case. As we approach the era of precision sub-percent cosmology, such biases are actually important, but Bernal et al. propose a methodology to move towards less biased BAO measurements and add more flexibility in the presence of models which can change the BAO feature in an important way. I really enjoyed reading this paper as it addresses a question I have always had in the back of my mind ever since at the start of my PhD I learned how BAO measurements are obtained. I think in general Bernal et al.’s paper is worth reading even if you are not interested in this question but simply want a concise but still correct summary of how BAO measurements are obtained (if this is the case, I strongly recommend reading Section III!).
#4 Bonus discussion for those who are interested
This is a bonus discussion I couldn’t resist writing after reading the paper by Bernal and collaborators. It is not my intention to be provocative, but just to state from my point of view what is the state of affairs with regards to the model-independence of BAO measurements. Above I wrote that this paper only partially addresses the BAO model-independence question. Let me elaborate on why I used the word partially. The paper focuses only on early-time deviations from ΛCDM as these are the ones which are argued to be most likely to solve the H0 tension. However, perhaps provocatively, I think it is definitely worth checking whether late-time deviations from ΛCDM can bias BAO measurements to a similar extent: else the inverse distance ladder argument that late-time solutions to the H0 tension are too constrained by BAO risks being circular. If late-time deviations from ΛCDM break one of the two assumptions outlined by Carter et al. in 1906.03035, one might in principle be concerned they might bias BAO measurements (Carter et al. tested specifically the dark energy equation of state and found possible 0.1% biases in the BAO parameters, which can be important moving on). In general, late-time deviations are much more likely to violate the first rather than the second assumption.
However, there is another paper I would recommend anyone interested in the issue of BAO model-independence reads, and it is 1908.11508 by Heinesen et al.: here, the authors (which include Chris Blake, an early force and expert in BAO measurements) quantify the accuracy of the Alcock-Paczynski (AP) scaling which lies at the heart of BAO measurements. They find a systematic budget of order the typical BAO statistical error due to failure of the AP scaling for models where large metric gradients are allowed at late times (this is important when taking into account the effect of structure formation, which is related to inhomogeneous cosmology, which is related to back-reaction, whose mentioning in turn opens a whole can of worms, etc.). This paper is extremely technical and I have not yet been able to digest it fully (I’ve been reading it for a couple of months now). However, I am just going to quote the following verbatim from the conclusions and leave it there: “Our analysis based on test cases indicate that the error budget in the standard literature is significantly underestimated when interpreting the herein measured distance scales as “model-independent” and using the results for constraining alternative models which are not close to the fiducial model. The additional systematics must either be included in the error budget, or alternatively it must be stressed that the results are not to be extrapolated to model cosmologies which differ more than a few percent from the fiducial model in terms of distance measures. It is also worth noting that the fiducial model is typically chosen to be close to a concordance model, which is in practice constrained from the CMB and other cosmological probes – and that caution must thus be taken about such implicit application of priors in BAO data reduction”.
Putative late-time solutions to the H0 tension will inevitably lead to distance measures which differ more than a few % from ΛCDM. If we are to believe this paper, then BAO measurements cannot be trusted to constrain these models (making the inverse distance ladder argument circular as I wrote earlier). I wrote to one of the authors of this paper some time ago to tell them I was really enjoyed it, and part of the answer was: “yes, I think that a late epoch explanation of the Hubble tension is likely. From my background in the field of Inhomogeneous cosmology I would say that we are potentially making large errors when forcing a global perfectly homogeneous and isotropic model onto a system with complicated structure.” I would suggest the interested reader also takes a look at 1712.02967 (another extremely technical read, but worth it if you have the time and interest). For the moment, I remain agnostic with regards to the issue, but I would really like to see someone expert in inhomogeneous cosmology repeat the analysis by Bernal and collaborators.