Probing Lyman-alpha Absorbers in Cosmological Simulations with Double Lines of Sight

Abstract

We perform a double line of sight (DLOS) analysis of the Lyman-alpha forest structures that form and evolve in cosmological N-body/hydrodynamic simulations. Pairs of simulated spectra, extracted from lines of sight separated by distances from D=12.5kpc up to 800kpc, and a ``control sample'' of unrelated lines of sight, are analyzed at redshifts 3, 2, and 1. Coincident line samples are defined for HI column density thresholds of Nco = 10^{12.5}, 10^{13}, and 10^{14} per square cm. We find that: 1) Under the assumption of a single structure size, a Bayesian analysis yields sizes that are larger for smaller Nco, and at fixed Nco the size decreases with decreasing redshift. However, these derived sizes are found to increase with increasing D indicating that the assumption of a single structure size is invalid. 2) The column densities of coincident pairs are highly correlated for small D, with increasing scatter as D is increased, consistent with structures that have a centrally peaked N(HI) that decreases gradually with radius. 3) The velocity difference distribution for coincident lines is very narrow for small D, and widens as D is increased to meet the expectation for chance coincidences in unrelated lines of sight. This behavior is indicative of organized motion within the structures. 4) For small D, the distribution of anticoincident line column densities, Nac, falls steeply as Nac increases from the cutoff value, but has a significant tail at large values which is inconsistent with a population of spherical absorbers with sharp edges, and consistent with a flattened geometry. The conclusions reached on the basis of the DLOS analysis are validated by an examination of the three-dimensional structures and velocity flows in the simulation data.