Speaker
Description
We know very little about the first galaxies that started the Cosmic Dawn (CD) and led to the Epoch of Reionization (EoR). Theory predicts these objects are highly biased, experiencing episodes of stochastic star-formation and feedback. As a consequence, radiation fields driving the CD and the EoR are expected to be very inhomogeneous. The inhomogeneity and timing of these cosmic milestones can be measured with CMB and other EoR and CD observations. We use efficient semi-numerical simulations to compute 3D lightcones of patchy EoR and other cosmic fields. We forward model the latest CMB and galaxy observations, including the patchy kinetic Sunyaev-Zeldovich signal, the Thomson scattering CMB optical depth and galaxy UV luminosity functions. We perform Bayesian inference, constraining the timing and duration of the EoR to be very late, and fairly rapid. We also find that the characteristic escape fraction of ionizing photons is $f_{\rm esc} = 4\%_{-3}^{+5}$. Additionally, we quantify the impact of different sources of stochasticity in determining galaxy emissivities during these epochs. As expected, stochasticity becomes increasingly important at earlier times when sources are rarer. We find that the star forming main sequence and the ionizing escape fraction are the most important sources of scatter in high redshift emissivities. Models seeking to accurately capture the morphology of the CD and EoR need to take this scatter into account. To constrain these models, we need to directly probe the morphology of the EoR. In order to do that we developed a new framework to infer the properties of the local HII region around a group of observed galaxies. We forward-model Lyman-alpha spectra as would be observed by JWST, dealing with all relevant sources of uncertainty. We find that the observed galaxy number densities of $n_{\rm gal} \sim 0.0025-0.00375$ cMpc$^{-3}$ should suffice to estimate the size and location of the local HII region. Such number densities are well within reach of future JWST surveys.
