Session 104 - IGM.
Display session, Thursday, January 16
Metropolitan Ballroom,

## [104.02] Probing the High Redshift IGM: SPH+P^3MG Simulations of the Lyman-\alpha Forest

J. Wadsley, J. R. Bond (CITA)

Our understanding of the Lyman-alpha forest has received a great boost with the advent of the Keck Telescope and large 3D hydrodynamical simulations. We simulate the high redshift universe using the SPH technique with a P^3MG (Particle-Particle Particle-MultiGrid) non-periodic gravity solver. We employ a high resolution (1 kpc) inner volume, essential for capturing the complex gas physics, larger medium and low resolution volumes surrounding it, essential for correct larger scale tidal fields, and a self-consistently applied, uniform tidal field to model the influence of ultra long waves. Such care is needed because the power per decade in the density fluctuations falls off very slowly in the dwarf galaxy regime of relevance to Lyman alpha clouds. The oft-used periodic boundary condition approach to simulations is ill-suited to proper treatment of the tides.

We use constrained field realizations to probe a selection of environments, including voids, quiescent regions, proto-dwarf galaxies and regions experiencing strong tides, such as large galaxy halos and galaxy-galaxy filamentary bridges. We statistically combine our simulations to provide a more comprehensive sample of the universe, including rare event'' regions which are difficult to obtain in unrestricted FFT-based approaches.

We fit Voigt profiles to the Lyman alpha spectra computed from our simulations direct comparison with the data, e.g., the column density distribution, line widths, temperatures, multiple line-of-sight correlations and the HI (and HeII) flux decrements. We demonstrate the importance of the photoionizing UV flux level and history, (2) tidal environment and (3) differing cosmologies, including CDM and CDM+Lambda. With galaxy-scale rms fluctuations \sim 1 at z=3 and a UV choice motivated by proximity effect observations, the simulations give results in excellent agreement with the data.