Cosmological radiative transfer for the line-of-sight proximity effect

Aims. We study the proximity effect in the Ly alpha forest around high redshift quasars as a function of redshift and environment employing a set of 3D continuum radiative transfer simulations. Methods. The analysis is based on dark-matter-only simulations at redshifts 3, 4, and 4.9 and, adopting an effective equation of state for the baryonic matter, we infer the HI densities and temperatures in the cosmological box. The UV background (UVB) and additional QSO radiation with Lyman limit flux of L(VLL) = 10(31) and 10(32) erg Hz(-1) s(-1) are implemented with a Monte Carlo continuum radiative transfer code until an equilibrium configuration is reached. We analyse 500 mock spectra originating at the position of the QSO in the most massive halo, in a random filament, and in a void. The proximity effect is studied using flux transmission statistics, in particular with the normalised optical depth xi = tau(eff),(QSO)/tau(eff),(Ly alpha), which is the ratio of the effective optical depth in the spectrum near the quasar to that in the average Ly alpha forest. Results. Beyond a radius of r > 1 Mpc h(-1) from the quasar, we measure a transmission profile consistent with geometric dilution of the QSO ionising radiation. A departure from geometric dilution is only seen when strong absorbers transverse the line-of-sight. The cosmic density distribution around the QSO causes a large scatter in the normalised optical depth. The scatter decreases with increasing redshift and increasing QSO luminosity. The mean proximity effect identified in the average signal over 500 lines of sight provides an average signal that is biased by random large-scale density enhancements on scales up to r approximate to 15 Mpc h(-1). The distribution of the proximity effect strength, a parameter that describes a shift in the transmission profile with respect to a fiducial profile, provides a measurement of the proximity effect along individual lines of sight. It shows a clear maximum almost without any environmental bias. This maximum can therefore be used as an unbiased estimate of the UVB. Different spectral energy distributions between the QSO and the UVB modify the profile but this can be reasonably well corrected analytically. A few Lyman limit systems have been identified that prevent the detection of the proximity effect because of shadowing.