The $\mathtt{WebSky}$ $\mathrm{[CII]}$ Forecasts and the search for primordial intermittent non-Gaussianity

Abstract

We present the $\mathtt{WebSky}$ $\mathrm{[CII]}$ line-intensity mock maps and forecast the capabilities of upcoming wide-field submillimeter-wave surveys of cosmological $\mathrm{[CII]}$ emission from the epoch of reionization (EoR). Using the $\mathtt{Peak~Patch}$ algorithm to generate light-cone dark matter (DM) halo catalogues and the $\mathtt{WebSky}$ framework to forward-model the cosmological $\mathrm{[CII]}$ signal, we construct tomographic mock surveys matched to the CCAT Observatory. We investigate both astrophysical models of $\mathrm{[CII]}$ emission from interstellar gas and the potential for the study of primordial intermittent non-Gaussianity (PING) as a science case for Stage 2 line intensity mapping (LIM) surveys. The $\mathrm{[CII]}$ voxel intensity distribution (VID) is used as a summary statistic in forecasts. Additional constraints on PING are derived from a relative entropy study of $\mathtt{Peak~Patch}$ halo mass functions. We show that upcoming LIM surveys will provide insights into the way we model cosmological line emission, and next-generation surveys can place competitive bounds on novel inflationary scenarios such as PING. The $\mathtt{WebSky}$ $\mathrm{[CII]}$ mocks and corresponding $\mathtt{Peak~Patch}$ halo catalogues are publicly available at https://uoft.me/webskycii .

Figures (14)

• Figure 1: The transfer function from from the inflaton-like field, $\Delta\phi_e$, to the asymptotic freeze out PING $\zeta$ response. The wavenumber, $k$, is expressed in units of the comoving Hubble scale at the end of the instability, $a_eH_e$. Modes, $\Delta\tilde{\zeta}_f$, with $k$ larger than about $k_{\mathrm{inst},p}/(a_eH_e)$, shown as the black dashed line, are subject to the $\chi$ smoothing and $T_{W\chi^2_e\to\Delta\phi_e}(k)$ transfer function which are low-pass filters, so high-$k$ behaviour of this transfer function will not affect $\zeta$.
• Figure 2: Possible PING features at $k_\mathrm{pulse}$ of $4\times10^{-4}$, $8\times10^{-3}$, and $1.6\times10^{-1} ~ h/\mathrm{Mpc}$ (in red, cyan and pink respectively), and amplitudes $\mathcal{D}_0$ of $2$, $2\times10^{-1/2}$, and $2\times10^{-1}$ (solid, dashed and dotted lines respectively) contrasted with the 2018 Planck matter power constraints with $1\sigma$ error bars as shown in Figure 19 of planckcollaboration_2018_I. The solid black curve shows the primordially Gaussian theory curve with best fit parameters from Planck using the CLASS Boltzmann solver. Tight constraints on the matter power at the peak of the power spectrum heavily rule out even a small PING effect, but further from the peak moderate PING could be possible.
• Figure 3: Posterior probability distributions generated using the Cobaya Markov chain Monte Carlo (MCMC) and CLASS Botlzmann solver codes. The solid blue and light blue shaded regions denote $1\sigma$ and $2\sigma$ joint posterior probability contours for a universe with a PING-like feature in the primordial power spectrum. Orange contour lines are the $1\sigma$ and $2\sigma$ joint posterior probability contours for a power law primordial power spectrum. The Posterior probability density function (PDF) for each parameter is shown along the diagonal with the mean and $1\sigma$ uncertainties listed above in colours corresponding to the contours. As suggested in Figure \ref{['fig:PING vs CMB matter power']}, posteriors disfavour $k_\mathrm{pulse}$ near the peak of the power spectrum and place an upper bound on the PING feature amplitude, $\mathcal{D}_0$. Cobaya and CLASS do not include moments beyond the two-point, so this is in effect a constraint on a primordially Gaussian universe with a PING-like feature, not a true constraint on PING.
• Figure 4: The PING scalar field, $\zeta$, is contrasted with a phase-randomized Gaussian field. The top row from left to right shows the Gaussian field, $\zeta_g$, the PING field, $\zeta$, and the response to the inflationary PING potential feature, $\Delta\zeta$, with $\mathcal{D}_0 = 0.6$ and $k_\mathrm{pulse}^{-1} = 5\times10^{-1/2} ~ \mathrm{Mpc}$. The half-wavelengths, $\pi k_\mathrm{pulse}^{-1}$, are shown as this corresponds roughly to the diameter of a PING peak. The bottom row from right to left shows phase-randomized perturbation, $\Delta\zeta^\mathrm{PR}$, the phase-randomized field added to the identical Gaussian field, $\zeta_g$, and the Gaussian field again. The black contours in $\Delta\zeta$ and $\Delta\zeta^\mathrm{PR}$ show $\pm2\sigma_{\zeta_g}, \pm3\sigma_{\zeta_g}, \pm4\sigma_{\zeta_g}$. The prominent peaks in $\Delta\zeta$ and absent in $\Delta\zeta^\mathrm{PR}$ show how power spectrum-only analyses of PING features cannot capture their complete NG effects.
• Figure 5: The probability density functions are compared for the PING response, $\Delta\zeta$, and the phase-randomized field, $\Delta\zeta^\mathrm{PR}$, as shown in Figure \ref{['fig:phase scrambled fields']}. Each PDF is measured from a $532^2$-voxel realization of the respective field. The dashed line is the best-fit Gaussian to the probability density function, $p(\Delta\zeta^\mathrm{PR})$. The phase-randomized field is Gaussian distributed, whereas the PING field is heavily skewed.