21-cm Background Anisotropies Can Discern Primordial Non-Gaussianity

TL;DR

The non-Gaussianity of initial perturbations provides information on the mechanism that generated primordial density fluctuations and 21-cm background anisotropies due to inhomogeneous neutral hydrogen distribution prior to reionization captures information on primordial non- Gaussianity better than a high-resolution cosmic microwave Background anisotropy map.

Abstract

The non-Gaussianity of initial perturbations provides information on the mechanism that generated primordial density fluctuations. The expected non-Gaussianity for slow-roll inflationary models is well below the ultimate detection level with cosmic microwave background (CMB) anisotropies or large-large structure at low redshifts. We find that 21-cm background anisotropies of the low-frequency radio sky due to inhomogeneous neutral Hydrogen distribution at redshifts between 30 and 100 captures information on primordial non-Gaussianity better than any other cosmological probe. An all-sky 21-cm experiment over the frequency range from 14 MHz to 40 MHz using a bandwidth of 1 MHz and with angular information out to a multipole of 10^5 can limit the primordial non-Gaussianity parameter f_nl < 0.01. The 21-cm background should eventually reveal the non-Gaussianity associated with single field slow-roll inflation.

Figures (1)

• Figure 1: (a) The 21-cm anisotropy power spectrum ($\sqrt{l^2C_l/2\pi}$; solid line) and non-Gaussian squared 21-cm anisotropy---21-cm anisotropy power spectrum ($[l^3X_l/2\pi]^{1/3}$; see text for details) with the optimal filter applied (see, equation 12). The dashed line is the signal related to the primordial bispectrum with $f_{\rm NL}=1$, while the dotted line is the noise bias $X_l^{\rm grav}$ from perturbation theory bispectrum under non-linear gravitational clustering. The dot-dashed line shows the total noise spectrum, $[l^3N_l/2\pi]^{1/3}$, associated with the measurement of $X_l^{\rm prim}$. These spectra are calculated at $z=100$. (b) The cumulative signal-to-noise ratio for a detection of the non-Gaussian spectrum $X_l$ as a function of the multipole $l$ at $z=100$ (solid line) and at $z=30$ (dashed line). These estimates assume no detector noise with measurements limited by cosmic variance alone and with $f_{\rm sky}=1$.