This week’s post is dedicated to dark energy in the context of (not solving) the Hubble tension, the possibility that the NANOGrav pulsar timing array may have detected non-tensorial gravitational wave polarizations, and the construction of a general covariant action for the so-called holographic dark energy model. Enjoy the read and have a nice weekend!
#1 2101.08641: Hockey-stick dark energy is not a solution to the H0 crisis by David Camarena and Valerio Marra
This week’s paper has many common features with two papers I discussed last year in my blog: 2002.11707 by Benevento et al. (see my 2020 Week 9 post) and 2001.09260 by Dhawan et al. (see my 2020 Week 5 post), which I’ll refer to respectively as B20 and D20 hereafter. Despite all the discussions about the fact that the Hubble tension likely calls for pre-recombination new physics (because of what the CMB+BAO+Pantheon combination tells us), naïvely one would think that a transition at very late times could still solve the Hubble tension. Since the Pantheon sample really only goes down to z~0.01, if the Universe sped up dramatically at z<0.01, surely this should raise H0 without conflicting with inverse distance ladder measurements, right? This possibility was first considered in detail in a 2009 paper by Mortonson et al., well before the Hubble tension was even a thing (the good old times, right?). In practice what one needs is a sudden discontinuity in the expansion rate of the Universe, which should start drastically increasing at a certain transition redshift. One way to achieve this is via a phantom transition, where the dark energy equation of state goes from a cosmological constant-like behaviour (w~-1) to a sudden deep phantom behaviour (w<<-1). A plot of the resulting w(z) would look a bit like a hockey stick (I’ll admit I had to Wikipedia this!), see below, hence the name hockey stick dark energy, or hsCDM.
Now of course this is naïve and ends up not solving anything, though the reason is not so obvious. It amounts essentially to a calibration mismatch on the Type Ia Supernovae (SNeIa) absolute magnitude M_B: the M_B which is behind the local H0 is incompatible with the M_B needed to fit the inverse distance ladder combination (CMB+BAO+Pantheon) and this remains true even for late-time transitions such as in the hsCDM model. Camarena and Marra then go on to make another very good point. It is in general very misleading to include a local H0 prior in one’s analyses. If one really wants to do this, one should instead include a prior on the local M_B, which carries at least three advantages over the use of a local H0 prior:
-it avoids double-counting the calibrator SNeIa sample, which contains almost 200 SNeIa in common with the Pantheon sample in the range 0.023<z<0.15;
-it avoids fixing the deceleration parameter to the standard value q0=-0.55, which strictly speaking should only be valid for LCDM or small deviations from it;
-it more consistently accounts for both the fact that M_B is constrained by the local calibration of H0, and the calibration mismatch between local H0 and inverse distance ladder;
Regular readers of my blog know that I am not a fan of including a local H0 prior anyway (though for a different reason than that discussed in this week’s paper, see my 2020 Week 26 post), but the reason for not doing so highlighted in this week’s paper is much more important: it is misleading to include the local H0 prior in one’s analyses even if by some miracle you manage to get a high H0 from the inverse distance ladder combination. Rather, one should properly re-calibrate the absolute magnitude of Pantheon SNeIa using the local distance ladder information, which now accounts for this higher H0. The best way to do this of course would be via a joint calibrator-Hubble flow SNeIa likelihood as the one developed in D20. As the authors, I “encourage the community to adopt […] the prior on the supernova absolute magnitude M_B as an alternative to the prior on H0.” (verbatim from the abstract) or, even better, don’t include any local information if you can help it. This (in my opinion rather subtle) point was essentially already present in B20, but the reason I enjoyed this paper by Camarena and Marra is that I believe they made the same point in a way which is cleaner and easier to digest. And, since the Hubble tension is a very hot topic which attracts many incorrect analyses,it is always good to remember the old Latin saying “repetita iuvant” (“repeating does good”), particularly in the context of subtle points such as the one discussed here. As a final note, to extract the prior to use on M_B given a certain local measurement of H0, you might want to check out the public code CalPriorSNIa which accompanies this week’s paper.
#2 2101.06869: Non-tensorial Gravitational Wave Background in NANOGrav 12.5-Year Data Set by Zu-Cheng Chen, Chen Yuan, and Qing-Guo Huang
In my blog I already discussed the exciting NANOGrav detection in my 2020 Week 38 post, where the possibility that NANOGrav might have detected cosmic strings was entertained. NANOGrav is a pulsar timing array (PTA) experiment aimed towards detecting a stochastic gravitational wave background (SGWB) at nHz frequencies by looking at the correlations between pulsar timing residuals. Sources of SWGB in this regime include merging supermassive black hole binaries, topological defects, and many others. In 2009.04496, the collaboration did report on the putative detection of a signal, although one should be very cautious about this possible detection. For one, the strength of the detected signal is in tension with previous upper limits, something which the collaboration attributes to different choices of priors in the statistical analysis. More importantly, no detection of spatial quadrupolar correlations has been found. This is something which is hard to reconcile with a genuine SGWB detection within General Relativity (GR). In short, NANOGrav has definitely seen correlations among pulsar timing residuals, but what these correlations are due to is presently unclear. Among the various proposed explanations, I entertained the rather crazy possibility that NANOGrav might have detected an inflationary SGWB in 2009.13432, recently published in MNRAS Letters. Let’s just say that this possibility cannot be completely excluded, but the resulting inflationary model and reheating dynamics would have to be baroque to say the least.
What does the absence of detection of spatial quadrupolar correlations imply? It essentially means that NANOGrav did not find evidence for tensor transverse (TT) correlations. Recall that in GR one expects two such transverse (and traceless) polarization modes for gravitational waves (GWs) associated to the massless spin-2 graviton. However, in theories of gravity beyond GR one can generally expect the presence of extra GW polarization modes: up to two extra vector modes and two extra scalar modes for the most general metric theory of gravity. These extra polarization modes would show up with different signatures compared to the TT signature in PTA experiments. Mathematically speaking, the resulting cross-power spectral density for the pulsar timing residuals is given by a sum over polarization modes of so-called overlap functions,weighted by appropriate coefficients. The overlap functions for different polarization modes are different, and the overlap function for TT correlations is known as the Hellings-Downs curve. In short, NANOGrav found that there is insufficient evidence for Hellings-Downs correlations in their detection.
In this week’s paper, Chen and collaborators then test whether the NANOGrav signal might be consistent with GW polarizations other than the TT ones. Possibilities include scalar transverse (ST), scalar longitudinal (SL), or vector longitudinal (VL) polarizations. Cutting to the chase, the authors re-analyze NANOGrav data and find that indeed the Bayesian evidence for TT modes (over pulsar-intrinsic red noise) is too low to claim a detection, consistently with what the NANOGrav collaboration finds. Same goes for SL and VL modes, for which the evidence is much lower. However, the evidence in favor of ST modes is extremely high. The results do depend on the assumed Solar System Ephemerides (SSE), for which various models exist. The authors test the DE438 and BAYESEPHEM SSE,and find rather drastic changes in the resulting Bayes factor in favor of ST modes - 99 vs 19 (!), see their Table 2 - which however remains always extremely high. If correct (and I stress that I am not enough an expert on these topics to judge their correctness), these results would suggest that NANOGrav might have detected scalar transverse GW modes, which can only arise beyond GR. One example of such a model would be the well-known f(R) model. It will be interesting to see if the NANOGrav collaboration will confirm or refute these results, and most importantly whether these hints will persist in future PTA data. And of course an interesting and related question might be whether such a strong detection of an extra polarization (but not the TT GR ones!) is contradicting independent tests of gravity which tell us that GR is faring so well.
#3 2101.08092: An effective field theory of holography dark energy by Chunshan Lin
Among the many theories of dark energy (DE) which have been proposed since the discovery that the expansion of the Universe is accelerating, holographic dark energy (HDE) is a particularly interesting one, and arises if one applies the holographic principleto the Universe as a whole. In a given region of the Universe, the largest length of a consistent QFT determines its infrared (IR) cutoff, and this cutoff should also be related to the ultraviolet (UV) cutoff, which instead determines the smallest length of this QFT. The zero-point energy of a system would appear to scale (and diverge) quartically with the UV cutoff. As the size of the system approaches that of a black hole (BH) whose Schwarzschild radius is given by the IR cutoff, this scaling can no longer hold since the total energy must be bounded by above by the BH mass, as the effective QFT would break down at smaller scales. This requirement effectively translates to an inequality which, if saturated, leads to a correspondence between the UV and IR cutoffs. The result is an effective DE-like component whose energy density scales as L^-2,with L the IR cutoff, and the UV cutoff essentially determining the magnitude of the effective vacuum energy. Identifying the IR cutoff with the future event horizon leads to accelerated expansion on a FLRW background, as shown inhep-th/0403127: thus the name holographic dark energy. HDE was first proposed in hep-th/9803132 as a possible way to reconcile the breakdown of QFT in large volumes, as required by the Bekenstein bound, with the success of local QFT in describing particle phenomenology, and a recent review can be found in 1612.00345.
Despite its apparent successes, so far HDE has mostly been a phenomenological model. In other words, a general covariant action which leads to HDE in the regime of interest for late-time acceleration is missing. In short, we are lacking a low-energy effective field theory (EFT) for HDE. An attempt to address this issue was put forward in 1210.0966, which however at the time of writing remains unpublished (not that this necessarily means anything, in my experience). In this week’s paper, building up on this previous work, Lin attempts to provide what he refers to as the missing piece for the HDE jigsaw puzzle, i.e. write down a consistent low-energy EFT for HDE and see from what theory it might originate. In covariantizing the action, Lin pays particular attention to making sure that the action is free of theoretical pathologies, such as ghost instabilities, which are removed by imposing appropriate symmetries. At the lowest dimensional operator level, the low-energy EFT for HDE is essentially given by Eq.(3.27). Lin finds that this EFT effectively (no pun intended) describes a massive gravity theory, with the UV-IR holographic correspondence arising from the energy scale where the massive graviton strongly couples to itself (i.e. the UV cut-off is inversely proportional to the graviton Compton wavelength). The resulting massive gravity theory is quite different from the conventional one though, as the massive graviton only has 3 polarizations (one helicity-0 mode and two helicity-2 modes) instead of the expected 5 (2*spin+1). This results from global Lorentz invariance being broken by the scalar field configuration. Lin speculates that there might in this sense be an analog to the Higgs mechanism, but the exact details need to be further explored. This is a very technical but interesting paper which, if correct, would represent an extremely important step forward in the study of holographic dark energy models.