This week’s entry features the return of the Hubble tension, the first discussion in this blog on the lensing anomaly (Alens tension), and for the second time a discussion on cross-correlations between ultra-high energy cosmic rays and large-scale structure (this time done on real data). As a totally random observation, all three papers discussed this week feature authors which already appeared earlier in this blog. Misha Ivanov already featured on Week 7 (#2), Julien Lesgourgues already featured on Week 11 (again #1!), Marc Kamionkowski already featured on Week 16 (#3), Avi Loeb already featured on Week 15 (#3), and Pavel Motloch already featured with another single-author paper on Week 17 (#3). And no, this is a totally random observation and I’m not trying to imply anything about these authors or the quality of their work (you might want to read my welcome post disclaimers once again), but rather you might start to see some interesting patterns regarding the type of works I am interested in. Enjoy!
#1 2005.10656: H0 tension or T0 tension? by Misha Ivanov, Yacine Ali-Haïmoud, and Julien Lesgourgues
It’s been almost two months since I last discussed solutions or attempts to solve the H0 tension (since my Week 14 post), and that’s basically because in my opinion nothing new happened in the field in the meantime if not rehashes of old ideas. This week’s paper doesn’t solve the tension but somewhat thinks out of the box and serves “as a reminder that the fundamental culprits of tensions are not necessarily any single parameter whose measurements differ between different datasets” (verbatim from Page 15 except with “dataset” -> “datasets”), and quite unsurprisingly Julien is involved. The paper really opened my mind and I thoroughly enjoyed reading it. Back to business, T0 is the monopole CMB temperature. This number essentially fixes the CMB photon number density and was measured to exquisite (and I mean it!) precision by COBE/FIRAS and other probes, where exquisite=0.02% precision! For this reason people usually assume that T0 is fixed by measurements of the CMB blackbody spectrum. But one can ask the question: can we instead measure T0 from the CMB temperature anisotropy spectrum? Or, in other words, what is the effect of T0 on the latter? As is well known, T0 is a proxy for the age of the Universe, much as H0. Therefore, if primary CMB anisotropies (i.e. anisotropies generated prior to last-scattering) were all there is, there would be a perfect geometrical degeneracy between T0 and H0 in the CMB alone. In other words, you can adjust T0 and H0 to keep the angular scale of the first peak (usually called thetas) fixed if you move along the degeneracy direction H0*T0^1.2 = constant, and there would be no hope of constraining them simultaneously.
Fortunately, as Ivanov and collaborators show in this week’s paper, this is not the case, because primary CMB anisotropies are not all there is. The so-called late-integrated Sachs-Wolfe effect (a source of secondary anisotropies driven by the time-variation of gravitational potentials in the dark energy dominated era) on large scales (although those scales are dominated by cosmic variance) and the effect of lensing on the CMB help breaking the H0-T0 degeneracy, at least partially, allowing one to measure the two simultaneously from CMB anisotropy data alone. Of course, both the resulting value of T0 and H0 come with substantially larger error bars (although the other parameters are mostly unaffected), and the resulting H0=70.5±2.2 km/s/Mpc is at face value compatible with the local measurement within better than 2 sigma. Prior to obtaining this limit, Ivanov and collaborators discuss in detail the optimal choice of parameters to vary when constraining T0 from CMB anisotropy data: in practice, some of the usual LCDM parameters should be rescaled by appropriate powers of T0, as the corresponding observables are sensitive to the baryon or dark matter densities per CMB photon (see the discussion on the top right column of Page 12 for the optimal parameter space to sample). This is a reminder that one should always think very carefully about what parameters best describe the observable one is considering, or equivalently what parameters/observables should be kept fixed when varying others (a discussion Julien has always been on top of).
Now, with the previous discussion in mind, one can choose to improve the amount of geometrical degeneracy breaking by either using the FIRAS prior on T0 to infer H0 (which is what people usually do, leading to H0 of about 67.3±0.6), or by using the local SH0ES prior on H0 to infer T0. Doing so one unsurprisingly recovers a value of T0 which is in strong tension with FIRAS. How strong is the tension? You guessed it, about 4 sigma. It is just a different way of recasting the 4 sigma H0 tension assuming LCDM. Of course, one can call upon BAO data to come to the rescue and break the geometrical degeneracy. And in this case, one finds H0 of about 67.9±0.5 (as expected from the inverse distance ladder arguments), and therefore the H0 tension remains, whereas T0 is consistent with FIRAS. Therefore, if we believe BAO data, the Hubble tension (or, as Ivanov and collaborators suggest to call it, Hubble-Penzias-Wilson tension) remains unsolved even if we relax the assumptions underlying the CMB monopole temperature. Nonetheless, this remains a very valuable paper, with its out-of-the-box thinking which I personally find enjoyable and stimulating.
#2 2005.08998: Lensing anomaly as a fingerprint of alternative scenarios to inflation by Guillem Domènech et al.
In this blog I’ve written a lot about the H0 tension, but I have never touched upon other milder tensions, such as the cosmic shear/sigma8 tension or the so-called lensing anomaly (or Alens tension). At the level of the CMB temperature anisotropy power spectrum (hereafter simply “CMB power spectrum” for the sake of conciseness), the effect of lensing is to smooth the higher-order acoustic peaks. With all other cosmological parameters known, one can estimate how much lensing one would expect, or in other words how smooth the peaks should be. It turns out, however, that Planck power spectrum measurements want a bit more lensing than that. This is the so-called lensing anomaly, or Alens tension. It comes from Alens, a purely phenomenological parameter first introduced by Calabrese et al. in 0803.2309 to artificially rescale the amplitude of lensing in the CMB power spectrum. Planck 2018 data wants Alens>1 with a significance between 2 and 3 sigma depending on the likelihood used, but it is worth recalling that the CMB lensing power spectrum estimated from the temperature 4-point function shows no such anomaly. It is unclear whether or not this is a real tension or a fluke (although more data = higher sky fraction seems to support the latter hypothesis, see 1910.00483), but it is worth noting that the lensing anomaly is intimately tied to the recent debate around whether Planck favors a spatially closed Universe rather than a flat one (I won’t go into this further, but if you want to find out more, you can read 1908.09139, 1911.02087, 2002.06892, or this Quanta Magazine piece). In any case, it is fair to say that there haven’t been many attempts to address the Alens tension, not even close to the amount of work which has gone into trying to address the H0 tension.
A possible solution to the Alens tension could be oscillatory features in the primordial power spectrum which, once transferred to the CMB power spectrum, could mimic extra lensing if their frequency is similar to the acoustic peak frequency. Oscillatory modulations could arise from many different physical scenarios, either in inflation or alternatives to inflation. One particularly interesting possibility in this sense is that of so-called primordial standard clocks (PSCs), see for instance 1809.02603 (by no means the first paper on the subject, but one of the ones which got more attention recently): these are due to the classical or quantum oscillation patterns of massive fields in time-dependent backgrounds. The resulting patterns record the evolution of the scale factor as a function of time, a(t), provided the field’s mass is constant, and hence for sufficiently massive field (which probe high energy scales) might help us distinguish inflation from alternative scenarios.
In this week’s paper, building up on previous work by Domènech and Kamionkowski in 1905.04323, Domènech and collaborators study both so-called sharp feature and PSC signals, which imprint oscillatory modulations in the primordial power spectrum, in an attempt to address the Alens tension. The work highlights the importance of k-dependence in the amplitude and frequency of the oscillations, without which it is impossible to address the tension. Even so, they find it hard to completely address the tension, which remains at a level slightly below 2 sigma. Nonetheless, one of the models they tested, the so-called T1 model parametrizing a feature due an oscillating massive scalar field in a power-law Universe, did seem particularly promising and could be worth looking into further. Taken at face value, the results for this model appear to be consistent with the feature being generated during a phase where the Universe was contracting, possibly during a matter bounce or radiation bounce (with an ekpyrotic scenario being excluded). This intriguing result highlights how the Alens tension might be a sign of scenarios alternative to inflation. While I too was intrigued, I personally suggest extreme caution in taking these results too literally, as future tests especially from large-scale structure data (where these features should also appear) are required to further support or discard this possibility. In closing, an interesting thing to do would be to test this model against the new more powerful CamSpec TTTEEE likelihood developed by George Efstathiou and Steven Gratton in 1910.00483 (which David Spergel jokingly called the post-Brexit CamSpec likelihood in the KICC 10th Anniversary Symposium), with larger sky fraction (=“more data”) which appears to support the possibility that the Alens tension might be a fluke as I discussed earlier, but is not yet public, as far as I know.
#3 2005.08782: Cross-correlating 2MRS galaxies with UHECR flux from Pierre Auger Observatory by Pavel Motloch
Two weeks ago, in my Week 19 post, I described a very interesting paper by Urban et al. which proposed to study the harmonic cross-correlation between ultra-high energy cosmic ray (UHECR) maps and maps of tracers of the large-scale structure (LSS) such as galaxies. This could help shed light on the properties of UHECRs, with energies of up to the EeV order (E=10^18!), which we have been observing since 1961, but whose origin, injection spectra, chemical composition, and redshift distribution remain largely unknown. Measuring the harmonic cross-correlation between UHECRs and tracers of the LSS helps quantify the coherence between the two maps as a function of angular scale, which is important in case a large number of UHECR sources trace the given LSS tracer. In this week’s paper, Motloch goes on and actually performs what to the best of my knowledge is the first measurements of cross-correlation between UHECRs and galaxies. More precisely, Motloch cross-correlates the flux map of UHECRs detected by the Pierre Auger Observatory (PAO) with a galaxy overdensity map from the 2MASS Redshift Survey (2MRS), a spectroscopic survey covering almost 91% of the sky.
Assuming a simple linear bias model between the UHECR-galaxy cross-power spectrum and the galaxy auto-power spectrum, Motloch then tries to constrain this bias from the measured cross-correlation which, of course, has relatively low signal-to-noise. It turns out that, due to noise, it is simpler to consider the excess cross-correlation, as defined in Eq.(14). Motloch finds a 1.8 sigma detection of non-zero bias, with best-fit value of b~0.01 (by eye this is hard to see in Fig. 1). This value of bias looks quite low, and it could be an indication of either strong deflection of UHECRs by magnetic fields (the earlier paper by Urban et al. already mentioned that correctly modelling the effects of magnetic fields would be crucial to obtain robust inferences about UHECR properties), or the fact that UHECRs might only trace a subset of the 2MRS galaxies (for instance starburst galaxies). The analysis then takes an intriguing turn when an interesting excess cross-correlation is found on degree angular scales (multipoles ell between 235 and 295) when considering UHECRs with energies above 52 EeV. However, it is hard to estimate the significance of this excess, which should definitely not be done by eye, and one should not forget to take the look elsewhere effect into account. So it is possible (if not, I would argue, likely) that this might just be a fluke. Overall, an interesting short paper which for the first time manages to measure an interesting hybrid cosmological-astrophysical cross-correlation!