Interpreting multi-wavelength observations of the epoch of reionization from next generation telescopes

Abstract

The Epoch of Reionization (EoR) is one of the least understood periods in the history of the universe. In this project, our goal was to explore the po tential of future multi-wavelength observations in understanding this epoch. We first explore the possibility of using the redshifted 158 µm line from singly ionized carbon (CII) atoms for this purpose. The 158 µm line emitted by the early sources of lights is expected to trace the star formation rate dur ing the EoR. We have attempted to constrain the minimum dark matter halo mass, that hosts the early photon sources, using the CII power spectrum from the EoR using an Artificial Neural Networking (ANN) based signal emulator and Bayesian Inference via a Markov Chain Monte Carlo (MCMC) frame work. We discovered that since the CII power spectrum is very featureless (the only feature it has is variation of its amplitude with the minimum host halo mass of the sources), one requires a very large training data set to train the emulator for it to be able to predict the signal power spectrum with an appreciable degree of accuracy. In addition, it was found that despite us ing a large training set, the emulator seems to have a tendency to produce degenerate data at certain points, especially towards the boundaries of our parameter space. This anomaly was discovered by using the MCMC algo rithm in tandem with the emulator and comparing the obtained estimates of the minimum host halo mass with the same obtained from MCMC runs in tandem with the simulator. We have further shown that the CII ⊗ 21cm cross-power spectrum has more features compared to the auto-power spectra of CII and thus would require a smaller training set to emulate the same. In addition, since this cross-power spectrum is dominated by the fluctuations in the HI 21cm field, this can potentially be used as a more robust statistic for constraining multi ple EoR parameters. We plan to explore this thoroughly in future and would like to explore this statistics potential to constrain reionization history along with other EoR parameters.