CMB B-modes from cosmic inflation​
The search for the primordial B-mode polarization signal (a curl-like pattern) in the CMB is one of the greatest challenges in cosmology. Detecting this signal would provide direct evidence of the primordial gravitational waves of quantum origin predicted by cosmic inflation (e.g., Starobinsky, Sov. Astron. Lett. 1983), a phase of ultra-rapid exponential expansion of the Universe that occurred a fraction of a second after the Big Bang. The amplitude of the primordial B-mode power spectrum, expressed through the tensor-to-scalar ratio, encodes information about the energy scale of inflation. Such a discovery would represent a major breakthrough in the quest for a quantum theory of gravity. Several concepts for next-generation CMB experiments are therefore being actively developed to detect primordial B-modes. However, their detection remains a tough challenge because the signal is extremely faint (below 50 nK temperature anisotropies) and is obscured by polarized Galactic foreground emission by several orders of magnitude.
​​​​
(Click on paper's title to access the corresponding publication)
​​​
Remazeilles, Dickinson, Eriksen, Wehus, MNRAS 458, 2032 (2016)
​​
In this paper, ​​we carried out the first direct comparison of different CMB satellite concepts (LiteBIRD, CORE, PIXIE, EPIC, and PRISM), quantifying their respective sensitivities for detecting primordial CMB B-modes in the presence of complex Galactic foregrounds. We implemented Commander (Eriksen et al., ApJ 2008), a Bayesian MCMC parametric fitting method with Gibbs sampling for component separation, on a series of simulations with varying foreground complexity (dust with two modified blackbody components, synchrotron curvature, AME polarization), to propagate foreground modelling errors into the estimation of the tensor-to-scalar ratio. For these different experiments, we computed the uncertainty on the tensor-to-scalar ratio due to residual Galactic foreground contamination. We also evaluated the impact of incorrect Galactic foreground modelling on the tensor-to-scalar ratio, showing that experiments with the narrowest frequency coverage may fail to obtain chi-square evidence for incorrect modelling and biased detections of the tensor-to-scalar ratio. These results are essential for guiding the optimization of future CMB experiment designs aimed at detecting primordial B-modes.
​​
​​​​​​
Exploring cosmic origins with CORE: B-mode component separation
Remazeilles et al. (100+ co-authors), JCAP 04, 023 (2018)​​
​​
As coordinator of the Foregrounds Working Group of the CORE Collaboration, I led this comprehensive paper (over 80 pages) on the problem of astrophysical foregrounds and component separation in the search for CMB B-mode polarization (primordial gravitational waves predicted by inflation), in the context of the proposed European space mission CORE, submitted to ESA in 2017. Using parametric methods (Commander, x-Forecast) as well as blind methods (SMICA, NILC) on detailed simulations of the polarized sky, we carried out the full component separation and likelihood analysis to recover the tensor-to-scalar ratio. We demonstrated that, for the baseline design of the CORE space mission, Galactic foregrounds can be controlled with the desired accuracy to reconstruct the primordial CMB B-mode power spectrum at both reionization and recombination scales, for a tensor-to-scalar ratio as low as r = 0.005. Although CORE was not selected by ESA, this detailed work has clarified many aspects of the foreground problem for B-modes, paving the way for future CMB B-mode experiments.
​​
​​​
Foreground separation and constraints on primordial gravitational waves with the PICO space mission
Aurlien, Remazeilles, Belkner, Carron, Delabrouille, Eriksen, Flauger, Fuskeland, Galloway, Górski, Hanany, Hensley, Hill, Lawrence, Pryke, van Engelen, Wehus, JCAP 06, 034 (2023)​​
​
In this paper, we evaluated the capability of the proposed NASA probe-scale mission PICO (Hanany et al., 2019) to detect or constrain the tensor-to-scalar ratio, using simulated full-sky observations across the 21 PICO frequency channels between 21 and 800 GHz, with five plausible foreground models of increasing complexity and input tensor-to-scalar ratios of r = 0 and r = 0.003. We implemented map-based component separation with NILC, along with the first post-separation, full-sky delensing of CMB B-modes in map space, achieving 78% delensing. For most foreground scenarios, PICO can place unprecedented 95% upper limits of r < 0.00013 to r < 0.00027, and recover an input signal of r = 0.003 with 18σ to 27σ significance. The study highlighted how removing low or high-frequency bands can weaken or bias constraints and demonstrated that leveraging many small sky patches across a large sky coverage helps identify such biases. We also showed that relying on Gaussian approximations of the likelihood at the lowest multipoles (â„“ ≤ 12) may underestimate uncertainties on the tensor-to-scalar ratio by ~30%.
​