Separating the memory of reionization from cosmology in the Ly$α$ forest power spectrum at the post-reionization era

Abstract

It has been recently shown that the astrophysics of reionization can be extracted from the Ly$α$ forest power spectrum by marginalizing the memory of reionization over cosmological information. This impact of cosmic reionization on the Ly$α$ forest power spectrum can survive cosmological time scales because cosmic reionization, which is inhomogeneous, and subsequent shocks from denser regions can heat the gas in low-density regions to $\sim 3\times10^4$ K and compress it to mean-density. Current approach of marginalization over the memory of reionization, however, is not only model-dependent, based on the assumption of a specific reionization model, but also computationally expensive. Here we propose a simple analytical template for the impact of cosmic reionization, thereby treating it as a broadband systematic to be marginalized over for Bayesian inference of cosmological information from the Ly$α$ forest in a model-independent manner. This template performs remarkably well with an error of $\leq 6 \%$ at large scales $k \approx 0.19$ Mpc$^{-1}$ where the effect of the memory of reionization is important, and reproduces the broadband effect of the memory of reionization in the Ly$α$ forest correlation function, as well as the expected bias of cosmological parameters due to this systematic. The template can successfully recover the morphology of forecast errors in cosmological parameter space as expected when assuming a specific reionization model for marginalization purposes, with a slight overestimation of tens of per cent for the forecast errors on the cosmological parameters. We further propose a similar template for this systematic on the Ly$α$ forest 1D power spectrum.