1. OWLS: OverWhelmingly Large Simulations The formation of galaxies and the evolution of the intergalactic medium

2. Outline Introduction to OWLS Radiative cooling Feedback from star formation Star formation histories Intragroup medium Gas accretion

3. Sales OWLS people BoothDalla VecchiaSpringelTheunsTornatoreWiersma BertoneCrainDuffyMcCarthyVan de VoortHaas

4. OWLS features LOFAR IBM Bluegene/L Cosmological (WMAP3), hydro (SPH) Modified Gadget III 2xN 3 particles, N = 512 for most Two sets: –L = 25 Mpc/h to z=2 –L = 100 Mpc/h to z=0 Runs repeated many times with varying physics/numerics

5. Video of the evolution of a massive galaxy down to z=2 3 Mpc/h

6. Zoom CDV, OWLS project

7. OWLS: New gastrophysics modules Star formation JS & Dalla Vecchia (2008) Galactic winds Dalla Vecchia & JS (2008) Radiative cooling Wiersma, JS, & Smith (2008) Chemodynamics Wiersma et al. AGN feedback Booth et al.

8. Radiative cooling (above 10^4 K) What is typically done: H and He including optically thin photo-ionization Metal cooling ignored or assuming CIE and solar relative abundances

9. Video of density dependence Wiersma, JS & Smith (2008)

10. Radiative cooling above 10^4 K Photo-ionization suppresses metal cooling  cooling rates decrease by up to an order of magnitude Relative abundance variations are important  cooling rates change by factors of a few Tables of cooling rates, element-by- element, including photo-ionization available Wiersma, Schaye & Smith, arXiv:0807.3748

11. Galactic winds Thermal feedback is quickly radiated away due to lack of resolution Solutions: –Kinetic feedback –Temporarily suppress cooling Most cosmological simulations employ the SPH code Gadget, which uses kinetic feedback Our kinetic feedback differs from that of Gadget: –Not hydrodynamically decoupled –Winds are local to the SF event

12. 1e12 M , face-on, gas density Dalla Vecchia & JS (2008) 45 kpc/h

13. 1e12 M , edge-on, gas density Dalla Vecchia & JS (2008) 45 kpc/h

14. 1e12 M , edge-on, gas pressure Dalla Vecchia & JS (2008) 45 kpc/h

15. Galactic winds Hydro drag determines outcome, gravity only indirectly important Low mass galaxies: wind drags lots of gas out to the IGM High mass galaxies: drag quenches wind  fountain Most popular existing prescription overestimates the energy in the outflow by orders of magnitude The details of wind implementations have grave consequences Dalla Vecchia & Schaye, 2008, MNRAS, 387, 1431

16. Lots of plots of SFR histories Most of these were flashed by…

17. Simulating galaxy statistics Cooling and feedback are crucial, SF law and structure of the ISM are not (Too) much freedom in implementation of galactic winds  use other constraints, e.g.: Metal distribution Gas profiles 

18. 1 1 0.1 Groups at z=0: Scaled entropy McCarthy et al.

19. 1 1 0.1 Groups at z=0: Scaled entropy McCarthy et al.

20. Groups at z=0 Massive galaxies reside in groups  detailed information about gaseous environment from X-ray observations at z=0 Highly sensitive to (metal) cooling and feedback Simulations can match detailed entropy, temperature, density and abundance profiles surprisingly well But it is a challenge to reproduce both the optical and X-ray properties of groups

21. How do galaxies get their gas? Classical picture: Gas-shock heated to the virial temperature, then cools onto disk Recent modifications: –Much of the gas falls in cold through filaments, particularly in low-mass galaxies –Efficient AGN feedback requires a hot halo –Galaxy bi-modality may be caused by transition from cold to hot accretion

22. Hot and cold accretion

23. Did not get to these slides…

24. Gas accretion - Conclusions Cold accretion fraction sensitive to definition Halo accretion: –Independent of subgrid physics –Hot fraction increases with mass and with decreasing redshift –Smooth transition from cold to hot Disk accretion: –Sensitive to subgrid physics –Cold accretion dominates at all masses unless it is stopped by feedback

25. Conclusions – 1/2 Some predictions from hydro simulations suffer from subgrid uncertainties (e.g. SSFRs, LFs), others are robust (e.g. accretion onto halos) Even when predictions are uncertain, hydro simulations can pinpoint the important physical processes, e.g. –Star formation laws are helpful but not constraining –Cooling can and must be done better –Freedom in feedback implementations is currently the bottleneck  need higher resolution and a better treatment of metal mixing “Realistic” simulations of the formation of –Individual high-z dwarfs are within reach –Massive galaxies are still far beyond the horizon Comparisons with galaxy surveys are too challenging and not always the most productive strategy

26. Conclusions – 2/2 Progress is most likely to come from studies of gas properties: –intergalactic, intra-group and intra-cluster media Available: hard X-ray profiles Needed: soft X-ray and UV at high (spectral) resolution –HI and CO structure of individual galaxies –QSO/GRB absorption spectra DANGERS (rant): –Many groups use (nearly) the same subgrid recipes –Insufficient awareness of models ingredients –Much more discussion about numerical accuracy (e.g. resolution and SPH vs grid) than subgrid uncertainties –Pressure to reproduce observations Subgrid variations are at least as important as convergence tests!