This week’s entry covers an optimal fully Bayesian CMB lensing reconstruction on SPTPol data, how to use line intensity mapping to study dark matter decay, and the impact of the newly determined deuterium fusion rates on Big Bang Nucleosynthesis. Enjoy!
#1 2012.01709: Optimal CMB Lensing Reconstruction and Parameter Estimation with SPTpol Data by Marius Millea et al. (SPTPol collaboration)
Reconstructing the gravitational lensing of the Cosmic Microwave Background (CMB) is important for at least two reasons. First, it gives (integrated) insight into how structure grew at late times, and is potentially sensitive to ingredients which might modify this structure growth, such as massive neutrinos, non-standard dark sector models, modifications of gravity, and so on. Second, on sufficiently small scales lensing converts E modes into B modes, and hence lensing is a contaminant to searches for primordial B modes: undoing the lensing effect then requires reconstructing the lensing potential to high accuracy. The “traditional” way the lensing potential is estimated is via the quadratic estimator (QE) first proposed by Hu and Okamoto in astro-ph/0111606. However, for instrumental noise below ∼5 µK arcmin, the QE becomes no longer minimum-variance, and therefore suboptimal. A number of alternative lensing reconstruction estimates have therefore been proposed in recent years, given that we are now approaching this resolution threshold. One interesting idea explored in the past years by Millea and collaborators is a full Bayesian reconstruction of temperature, lensing potential, and polarization maps, alongside cosmological parameters. Inclusion of the latter was the key point towards making this method realistic. In practice, then, one samples a huge parameter space made up of the pixels of the aforementioned maps alongside cosmological parameters. This is a huge task, made possible to perform on incredibly short timescales, among the other things, by Hamiltonian Monte Carlo and GPUs. The paper which described the underlying methodology is Millea, Anderes & Wandelt, 2002.00965.
This week’s paper led by Millea uses data from the SPTPol receiver on the South Pole Telescope to perform the first simultaneous Bayesian parameter inference and optimal reconstruction of the gravitational lensing of the CMB, thereby demonstrating that the method can work in practice on real data. The SPTPol data has a noise level of 5.8 µK arcmin, so it is already reaching the level where the QE is suboptimal. There are a three key novel aspects of this work, in particular it is the first time that:
1) a parameter (Aφ, which controls the amplitude of the gravitational lensing potential) is estimated from an optimal lensing reconstruction;
2) parameters controlling the lensing potential and CMB bandpowers are jointly estimated while self-consistently accounting for the correlation between reconstruction and delensed CMB;
3) a fully Bayesian delensing method is demonstrated to work on real data.
This is an extremely important result as such methods will be more than warranted if future CMB experiments such as SPT-3G, Simons Observatory, and CMB-S4 will indeed reach their expected noise levels. Finally, this week’s paper also comes with a public code (thumbs up!), written in Julia and available here. This is one of the first times I have seen a major code in my field being written in Julia, something which is definitely making me feel a tiny bit not in line with the times!
#2 2012.00771: Strategies to Detect Dark-Matter Decays with Line-Intensity Mapping by José Luis Bernal, Andrea Caputo, and Marc Kamionkowski
Line intensity mapping (LIM) is the idea by which you target a specific emission line, such as HI, CO, CII, Hα, Lyα, or many others. Your observable is then the integrated intensity of this line at a given target frequency. Since you know the original frequency of the line, your target frequency immediately tells you what redshift you are observing. LIM is a promising way to characterize dark matter (DM), via the influence of the latter on the surrounding medium, for instance via exotic energy injection which, by heating and ionizing the intergalactic medium, will alter the intensity and fluctuations of the resulting line. However, LIM is of course not immune to contaminations and systematics. In particular, so-called interlopers consist of lines other than one’s target line, which of course contaminate one’s observations. And, of course, astrophysical uncertainties have to be taken into account and correctly marginalized over.
Compared to previous studies on using LIM to characterize DM, this paper has a different twist. In particular, the question the authors ask themselves is whether one can use LIM to directly observe the electromagnetic radiation produced by DM decays, rather than relying on the indirect impact of the latter on the intergalactic medium, which in turn affects the observed line. As an example, if DM is cold and annihilates to a 2-body final state with at least one photon, an emission line is produced. This extra emission will pretty much be an interloper, and would usually be treated a background. But since there are various techniques to characterize and extract interlopers, what is being proposed here is basically to treat the interloper as a signal, and extract information on DM from it. The authors then go on to show that this can work in practice, focusing on the LIM power spectrum and voxel intensity distribution (a voxel is the 3D analogue of a pixel), and showing that the latter will be more sensitive to the former. They find that one may be sensitive to DM of mass ~10^-6 - 10 eV decaying into two photons, and that for the particular case of the axion, this technique might be highly complementary to laboratory and astrophysical searches, allowing one to also probe part of the parameter space of the Peccei-Quinn axion. Overall, this was an interesting and detailed study showing that LIM holds great promise to detect the products of radiative DM decay, and thereby possibly shedding more light on the nature of the DM. It also comes with a public code (another thumbs up!) to compute various statistics of line intensity maps using a wide variety of models, available here.
#3 2011.13874: The Impact of New d(p,γ)He3 Rates on Big Bang Nucleosynthesis by Tsung-Han Yeh, Keith Olive, and Brian Fields
In my Week 46 entry I covered a key development in the field of Big Bang Nucleosynthesis (BBN), that is, an improved determination of the rate of Deuterium (d) fusion into Helium-3 (3He) upon absorption of a proton, a reaction usually referred to as d(p,γ)3He. This was made possible thanks to the efforts of the LUNA collaboration, which by observing the corresponding process under the Gran Sasso mountain, managed to minimize contamination from cosmic rays. In turn, this measurement has sharpened our understanding of BBN, and in particular our theoretical uncertainties on the primordial abundance of light elements, including d and 4He, which are particularly sensitive to the baryon density of the Universe, Ωbh2. In this respect, the purpose of this week’s paper is threefold:
1) to re-evaluate the thermal rate for d(p,γ)3He, using a world average of cross section data;
2) to re-evaluate the consistency of BBN, CMB, and observed light element abundances;
3) to understand where improvements are needed for further progress in the field.
In relation to the first point I mentioned above, the paper finds good agreement between their re-evaluation of the thermal rate for d(p,γ)3He and the LUNA results. More importantly still, excellent consistency is found between BBN theory, observed CMB anisotropies, and observed abundances of d and 3He (and therefore also inferences of Ωbh2 and Neff), with the consistency improving compared to the past. Interestingly, the observed abundance of d is more precise than the theoretical prediction. In other words, we are in a seldom seen situation where it is the theory that has to catch up with the data and not the other way round. In particular, the BBN+CMB prediction for the d abundance is now dominated by the uncertainty in the d(d, n)3He and d(d, p)3H reactions. I don’t know for sure what experiment is best suited to improve the determination of these rates, but the nice thing is that both reactions have the same initial state, and hence the rates should be accessible to the same experiment. Unless we are able to improve the theoretical BBN predictions by the time future CMB experiments such as CMB-S4 start taking data, comparisons between CMB and BBN data especially with regards to Neff will be limited by this theoretical uncertainty bottleneck which we are currently hitting. And, of course, the elephant in the room as far as BBN theory (or observations of light element abundances) is concerned remains, in the form of the well-known and currently unsolved Lithium problem.