SPH Cosmological Simulations of Galaxy Formation

I'm using several Gravity + Smoothed Particle Hydrodynamic (SPH) cosmological simulations of various box sizes  and resolutions  to study the formation and evolution of galaxies. SPH has several practical advantages over Eulerian approach (and also several disadvantages): it is naturally adaptive and it is Lagrangian in nature, which means that particles (i.e. fluid elements) can be followed both in space and time, enabling us easier study of processes like accretion of gas by galaxies.

Most of the results for my thesis are from the PtreeSPH code (Dave, Dubinski & Hrnquist 1997). Such simulations determine gravitational forces using hierarchical-tree algorithm while hydrodynamic forces are calculated by averaging i.e. smoothing the gas properties over the nearby fluid elements (particles).

Details of these simulations are described in Katz, Weinberg and Hernquist (1996). Example from one of these simulations made in TIPSY  is shown on the left (22 Mpc comoving box, z=2.25 , only filamentary and dense gas is shown).

I'm also extensively using GADGET-2 code from Volker Springel and the Gasoline simulations (Wadsley, Quinn & Stadel 2004).

All three of the above simulations codes, typically include UV background, star formation prescription and some sort of feedback.

Gas accretion in simulations: COLD mode vs. HOT mode

    It was shown earlier (Murali et al. 2002) that gas accretion is dominant process of growth of galaxies. Merging starts to be important only at low redshifts, while gas accretion is important at all times. In the same paper it was also shown that global Star Formation Rate (SFR) tightly follows gas accretion and not merging .  Therefore we need to study gas accretion in detail in order to explain global properties of galaxies today.

    In standard paradigm of galaxy formation (based on White and Rees 1978),  gas is accreted onto galactic disc in two stages:

1) gas is shock heated to the virial temperature of the parent halo, close to the virial radius ("virial shock") and then
2) shock heated gas in the halos radiatively cools and collapses onto galaxies.

    However, from simulations we see that bulk of the gas enters galaxies without ever being significantly shock heated and part of the accreted gas is heated much closer to the galactic disk than standard model predicts. This could have significant consequences on the scenario of galaxy formation.

    Results from our simulations show that history of gas accretion is highly different than standard paradigm of galaxy formation predicts. Gas can enter galaxies trough two modes: Cold mode with T_vir~10^4-10^5 K and "standard" Hot mode with T_vir~10^6K.
At high redshifts gas is rarely heated to virial temperature before it enters galaxies.  Significant fraction of gas stays "cold" even when it enters galactic halos and galaxies.  Only in the largest galaxies gas gets shock heated close to the virial temperature.
At intermediate redshifts comparable amount of hot and cold gas enters galaxies.
At low redshifts standard picture seems to be more valid, and most of the accreted gas is heated to T~T_{vir}. (Even here situation is more complicated since typically T is smaller than T_{vir} around the virial radius and it reaches T=> T_{vir} deeply inside the halo).

    Accretion of cold gas component goes mainly through filamentary accretion. In small halos ( <~ 2e11M_sun, for zero metallicity) almost all accreted gas is cold, especially at high redshifts. Filaments are able to penetrate deep into halos even at moderate redshifts and even in the case of moderate galactic masses where shock front is developing close to R_{vir}.

All of the statements above depend on the galactic mass and environment. Cold mode is more important in low mass galaxies and low density environments, while hot mode dominates in massive galaxies and dense environments.
In support of these results from SPH simulations is recent analytical work and simplified simulations of Birnboim and Dekel (2003) who showed that in the halos smaller than M_{halo} ~1e11 M_sun, virial shock does not develop.

Various dependencies of cold and hot mode of gas accretion are studied in detail. We also used simulations of wide range of resolutions to check resolution effects on our results. Our first preliminary results (august 2002) can be found here.
Recent detailed paper about cold and hot mode accretion in cosmological simulations is published in MNRAS and it can be found here (high resolution version) or  on ArXiv.org (low resolution).

Follow the link if you want to download some nice figures illustrating cold and hot mode accretion

Project Collaborators:

Neal Katz, U. of Massachusetts
David Weinberg, Ohio State U.
Romeel Dave, U. of Arizona

Useful tools for SPH simulation:

SKID
SMOOTH
TIPSY
 

Last modified: Dec. 2005
Author: Dusan Keres