Robust forecasts on fundamental physics from the foreground-obscured, gravitationally-lensed CMB polarization

TL;DR

Foreground contamination and lensing from large-scale structure significantly impede detection of inflationary B modes in the CMB. The paper presents a self-consistent forecasting framework that combines parametric maximum-likelihood foreground removal, iterative and cross-correlation delensing (CMB×CMB, CMB×CIB, CMB×LSS), and Fisher-matrix cosmological forecasts across a wide array of pre- and post-2020 instruments, with an online interface. It demonstrates that complementary ground, balloon, and space data can substantially improve component separation, delensing, and cosmological constraints, e.g., post-2020 configurations achieving $ \\sigma(r) \\sim 1.3\\times 10^{-4}$ and tight probes of $M_\\nu$, $N_{\\rm eff}$, $\\Omega_{\\rm k}$, and $w$ while controlling foreground residuals to $r_{\\rm eff} \\lesssim 10^{-4}$ and delensing contextual performance. The results highlight the synergy between Stage-IV, space missions, and foreground monitors (Planck, C-BASS, QUIJOTE-CMB) and provide a publicly accessible tool for optimizing frequency coverage and cross-experiment collaboration. This framework thus enables robust forecasts for inflationary and late-time physics from foreground-obscured, gravitationally-lensed CMB polarization.

Abstract

[Abridged] Recent results from the BICEP, Keck Array and Planck Collaborations demonstrate that Galactic foregrounds are an unavoidable obstacle in the search for evidence of inflationary gravitational waves in the cosmic microwave background (CMB) polarization. Beyond the foregrounds, the effect of lensing by intervening large-scale structure further obscures all but the strongest inflationary signals permitted by current data. With a plethora of ongoing and upcoming experiments aiming to measure these signatures, careful and self-consistent consideration of experiments' foreground- and lensing-removal capabilities is critical in obtaining credible forecasts of their performance. We investigate the capabilities of instruments such as Advanced ACTPol, BICEP3 and Keck Array, CLASS, EBEX10K, PIPER, Simons Array, SPT-3G and SPIDER, and projects as COrE+, LiteBIRD-ext, PIXIE and Stage IV, to clean contamination due to polarized synchrotron and dust from raw multi-frequency data, and remove lensing from the resulting co-added CMB maps (either using iterative CMB-only techniques or through cross-correlation with external data). Incorporating these effects, we present forecasts for the constraining power of these experiments in terms of inflationary physics, the neutrino sector, and dark energy parameters. Made publicly available through an online interface, this tool enables the next generation of CMB experiments to foreground-proof their designs, optimize their frequency coverage to maximize scientific output, and determine where cross-experimental collaboration would be most beneficial. We find that analyzing data from ground, balloon and space instruments in complementary combinations can significantly improve component separation performance, delensing, and cosmological constraints over individual datasets.

Figures (15)

• Figure 1: Schematic of the adopted methodology described in Sect. \ref{['sec:methodology']}.
• Figure 2: Left panel: Angular power spectra showing primordial $B$ modes, lensing $B$ modes, total intensity, and $E$ modes, as well as the total contribution of polarized $B$-mode foregrounds (dust plus synchrotron), expected on the cleanest 1--90% of the sky, at $100$ and $200$ GHz. Note that, as these results are derived from Planck's Galactic masks and are not therefore optimized for high-resolution, ground-based instruments, there is potential for discovery of small patches of sky (e.g., $f_{\rm sky} \lesssim 5\%$) cleaner than those indicated here. Right panel: The ratio of power spectra of foreground and lensing $B$ modes to primordial $B$ modes, assuming a tensor-to-scalar ratio $r=1$. The contours indicate, in effective values of $r$, the contamination due to foregrounds and lensing on primordial $B$-mode measurements. The $x$- and $y$-axes correspond to the multipole $\ell$ and frequency of observation, in GHz, respectively. The level of input foregrounds are estimated on a 50% patch of the sky.
• Figure 3: Left panel: the noise in the final CMB map for an experiment with two frequency channels of varying polarization sensitivity centered on 150 and 220 GHz, combined with Planck's 353 GHz channel. The sole foreground contaminant is dust. Right panel: the noise degradation factor $\Delta$ between the final CMB map and the quadratic combination of all channels for the same experimental configuration.
• Figure 4: Left panel: the delensing improvement factor, $\alpha$ (defined in Eq. \ref{['eq:alpha_def']}), for iterative CMB$\times$CMB (dark orange), CMB$\times$CIB (light orange) and CMB$\times$LSS (dark red) delensing, as a function of the noise in the CMB map after component separation. We assume FWHM=$3'$ and a multipole range of $\sqrt{ 2\pi / f_{\rm sky}} \le \ell \le 3000$. Figure \ref{['fig:sigma_cmb_example']} and its accompanying discussion describe three-frequency experimental configurations producing comparable noise levels after the removal of dust. Right panel: One-sigma limit on $r=0$, $\sigma(r=0)$, as a function of both the noise in the CMB map and the observed fraction of the sky. Grey (black) contours correspond to the no delensing (CMB $\times$ CMB delensing) case. No foreground residuals are taken into account in this result.
• Figure 5: Illustration of the frequency and multipole coverage of the instruments detailed in Sect. \ref{['ssec:missions']} and Appendix \ref{['app:instruments_specifications']}. The $\ell_{\rm max}$ values plotted indicate the multipole at which the noise and beam dominate all cosmological signal (where this is lower than the maximum multipole considered in this work, 4000).