1. Mutual feedback between star formation and nuclear activity at high z Implications for Galaxy Formation Models Gian Luigi Granato INAF and SISSA + L. Silva, A. Lapi, F. Shankar, L. Danese, G. De Zotti, A. Bressan, J. Mao

2. A model for the joint formation and evolution of spheroidal galaxies and high redshift QSOs Gian Luigi Granato INAF and SISSA + L. Silva, A. Lapi, F. Shankar, L. Danese, G. De Zotti, A. Bressan, J. Mao

3. Uncertainties in galaxy formation models in cosmological context For many years uncertainties both in cosmology and physical processes Now “precision cosmology era”, background model rather well constrained (though mysterious in nature) concordance  CDM model: (  m,  ,  b,h,n s,  8 ) ' (0.3,0.7,0.04,0.7,0.95,0.75 ¥ 0.9) Main physical processes driving luminous galaxy formation are extremely complex and still hotly debated

4. The problem with luminous matter To compare scenarios of galaxy formation with observations two very critical and uncertain steps: 1. Model the non linear evolution of baryonic matter: most driving processes occur well below the resolution of any simulation (sub-grid physics) and are also poorly understood. 2. Model the interaction of photons produced by stars and accretion processes with the dusty ISM. Come to Laura Silva talk next Tuesday!

5. Di Matteo et al 05 Simulation of merging of spirals without treatment of induced QSO activity…..and WITH (crude and uncertain sub-grid) treatment of induced QSO activity and ensuing feedback on ISM

6. Fate of initial gasWithout AGNWith AGN In stars89%52% Cold SF gas1.2%0% Hot halo gas9.8%11.1% Expelled from halo0.05%38% In SMBH-1.6% not The predicted star formation histories and final morphologies are completely different. This result is driven by processes not simulated from first principles. Di Matteo et al 05

7. Ab-initio models or toy models? First principles models do not exist: in any computation sub-grid physics is dominant and treated (if ever) through semi-analytic-like recipes Extensive comparisons between different scenarios and data are done by means of fully semi-analytic models (SAM) for baryons, possibly as post- processing of gravity-only simulations for DM. “By definition” of SAM, many a-priori assumptions on the general behaviour of the system. More proper naming could be toy models?.

8. Standard SAMs assume Most SAMs assume a disk galaxy merger driven sequence of processes leading to present day galaxy populations 1. The outcome of gas cooling in DMH is gaseous rotation supported disk, with mild SF (Rees & Ostriker 1977, Silk 1977, White & Rees 1978….) ; 2. Disk mergers are the only driver of bursty SF and of the formation of spheroids (White & Rees 1991, Cole 1991, … omissis…Cole et al 2000).

9. Problems of standard SAMs Calculations based on this general scheme show severe mismatches with observations –Difficult to reproduce the bright end of LLF. Models predict too many (and too young-blue) bright galaxies (e.g. Benson et al 2003) –Beyond z ' 1-1.5 too few bright galaxies –Cosmic downsizing –Absence of cooling flows in real rich clusters –Properties of local E galaxies Likely all facets of the same problem. Some key ingredient is missing or/and the scheme needs revision

10. Promising solution: mutual link between SF and AGN activity Strong and increasing observational evidences and theoretical suggestions of a link, e.g.: –M BH -spheroid relations (L sph, M sph,  sph ) –Similarity of cosmic SFR(z) and L QSO (z) –Simulated galaxy mergers drive gas to the centre Despite this, only very recently began to be incorporated into SAM.

11. Since M sph /M BH ~1000, feed-back from a SMBH could easily exceed the binding energy of the spheroid: Energy budget of AGN and SNae during spheroid growth mostly released on a short timescale (a few e-folding) To compare with released more “gently”

12. AGN feedback mechanism However the mechanism for energy injection is unclear –Radiation pressure, mostly on dust (Voit et al 1993, Murray et al 2005) –Radiative heating (e.g. Ostriker & Ciotti 2004) –Kinetic outflow from AGN (likely generated by RP on lines; e.g. Murray et al 1995) –Complex interplay between SN and AGN FB (Monaco 2004)

13. QSO mode vs Radio mode AGN feedback in SAM AGN feedback has been considered only very recently in SAMs, and in two well distinct flavours, with different aims: –FB associated with the main phase of BH growth, related to the bright high-z QSOs, to sterilize massive high-z galaxies, little affected by SNae (Granato et al 2001, 2004, Monaco & Fontanot 2005; Menci et al 2006) –FB associated with lower redshift, low accretion rate phase of AGN, optically irrelevant, to halt cooling flows and avoid overproduction of local bright galaxies (Bower et al 2006, Croton et al 2006, see also Cattaneo et al 2006)

14. ABC model (Granato et al 2001,2004; Silva et al 2005; Lapi et al 2006) Observations suggest an early and fast ( » 1 Gyr) assembly of most baryons in Es with more massive objects forming faster. To get downsizing within hierarchical assembly of DM we propose a revision of SAMs (Anti-hierarchical Baryonic Collapse ABC): 1. Reduce role of gas disk formation at high z: cool collapsing gas in big halos at high-z start vigorous SF without setting in a quiescent disk. 2. Large SFR promotes the development of SMBH from a seed, which after » 0.5 Gyr powers an high-z QSO. 3. Keep into account the feed-back of this QSO on the host.

15. ABC evolutionary sequence SMG – dusty ERO with growing SMBH High redshift QSO Red and dead massive high z galaxy Local Spheroid (E+bulges) with dormant SMBH » 0.5 Gyr » 0.05 Gyr » many Gyr The model passes all tests against these populations

16. Halos form, gas is heated to virial T Scheme of ABC (Granato et al 2004) at high z Gas cools, collapse and forms stars directly, in small halos SNae quench SF, in big ones nothing prevents a huge burst of SF ( ' 1000 M ¯ /yr over 0.5 Gyr), SMGs phase… (almost) passive evolution of stellar population. Red and dead massive galaxies at high z (ERO) with dormant SMBH..with SMBH growth promoted by SF eventually powering high z QSO after » 0.5 Gyr, which cleans ISM and quenchs further SF and then itself. QSO phase

17. HOT GAS COLD GAS RESERVOIR (low J) STARS IGM SMBH-QSO SNae feedback & QSO feedback Radiative cooling Radiation drag (  SFR) Viscous accretion Collapse baryonic components and mass transfer processes Stellar evolution Arrows give a set of simple differential equations for the masses in the various components, solved numerically

18. From cold gas to low J reservoir: radiation drag (  SFR) (Umemura et al…)

19. From cold gas to IGM: SNae Feedback From cold & hot gas to IGM: QSO Feedback (BAL),

20. Galaxy SMBH Accretion rate SFR VERY Dusty and huge SF ) SMGs – dusty ERO SMBH cleans the ISM ) high z QSO Little ISM, almost passive evolution ) passive ERO Local Ell and SMBH Plugging this into statistic of dark matter halos as a function of M vir and z vir we get predictions for many populations, connected by evolutionary sequence

21. Galaxy SMBH SFR Accretion rate Phase 1: VERY Dusty and huge SF and baby SMBH growth lasting » 0.5-1 Gyr ) SMG with mild obscured AGN activity – dusty ERO

22. ABC naturally reproduces SMGs (e.g. no ad-hoc IMF) 5.7 mJy z distMEDIANQUARTILE Chapman et al 2005 (73 sources) 2.21.7-2.8 Model2.21.6-3.3 SCUBA 850  m MAMBO 1200  m model data Silva et al 2005

23. THE PRE-QSO PHASE IN SMGs The build up by accretion of the SMBH gives rise to a mild AGN activity in sub-mm galaxies, detectable only in hard-X. This has now been found (Alexander et al 03,04,05) dM/dt(BH)>0.013 M ¯ /yr ) L(0.5-8)>1E43 erg/s dM/dt(BH)>0.13 M ¯ /yr ) L(0.5-8)>1E44 erg/s (Granato et al 2006)

24. By converse, the normal disk merging scenario for SMGs predicts too high M and dM/dt for the SMBH in SMG, because of the » 1 Gyr phase of disturbance and SMBG growth preceding the final merge and huge SF.

25. Phase 2: SMBH cleans the ISM ) high z QSO ( ' 5 £ 10 7 -10 8 yrs) Galaxy SMBH Accretion rate SFR T delay ' 0.3-1 Gyr, a key built-in feature

26. General methodology of “QSO only models” (e.g. Whythe & Loeb 2003; Mahmood et al 2004): Ingredients and assumptions: Haloes formation rate (e.g. derivative of HMF (PS) ) SMBH-DMH mass relationship (e.g. from self-regulation) SMBH mass-luminosity relation (e.g. Eddington) QSO appears immediately and shines for time t Q But the required shine time t Q is too short ( ' galaxy dynamical time-scale ' 10 7 ys at z=3 and a few 10 6 at z=6) to satisfy the Soltan argument and the local SMBH mass function with plausible accretion efficiency 10-15%.

27. In our view the problem is mainly due to neglecting the time delay between virialization and detectable QSO activity Lapi et al in 2006 Lapi et al 2006 HMF or formation rate has to be considered at z(t vir -t delay ) rather than at z(t vir ) ) less objects ) higher t Q Our ABC model for QSO-spheroids co-evolution has this delay and the intrinsic light curve built-in (but not the visibility time t Q ). The (high-z) QSO LFs are another fundamental test!

28. Optical QSO LF (t Q ' 4x10 7 yr) z=1.5 data Croom et al 2004 z=3 data Pei et al 1995 z=4.5z=6 data Fan et al 2004 Lapi et al 2006

29. X-ray QSO LF (t Q ' 10 8 yr) z=1.5z=2  Barger et al. (2005)  Ueda et al. (2003)  La Franca et al. (2005) Lapi et al 2006

30. The ABC model well reproduces the evolution of high-z optical and X-ray LF of QSO with the only addition of a plausible visibility time ' (a few) 10 7 yr in optical and ' 10 8 yr in X, as suggested by demography studies of local SMBH (Shankar et al 2005, Marconi et al 2005). The delay between halo formation and peak of SMBH accretion is a crucial ingredient.

31. Galaxy SMBH Accretion rate SFR Phase 3: Little ISM ' passive evolution ) red and (almost) dead massive high-z galaxies (many Gyrs)

32. Z ' 0.5 Z ' 0.9 Z ' 1.3 Z ' 1.8 Fontana et al 2004: galaxy stellar mass function in K20 sample Standard SAMs Granato et al 2004 Standard SAMs underproduce massive galaxies, by a fraction increasing with z

33. Adapted from Drory et al 2005 Massive galaxies at high redshift Baugh et al 2005 (Durham SAM) Granato et al 2004 ABC

34. Galaxy SMBH Accretion rate SFR Phase 4: Local Ellipticals and dormant SMBHs

35. Local K band Luminosity function of spheroids Data: Huang et al 2003 Kochanek et al 2001 Granato et al 2004

36. Sheroid-SMBH correlations  = 0.57 ± 0.05 V vir dispersion interpreted as different virialization epochs Tighter M BH -M * ?

37. Mass function of local SMBHs observations model

38. Passive Ellipticals All Ellipticals Late type Silva et al 2004, 2005 K Band counts

39. Onset of SF in early universe: (I) LFs of LBGs… Adopting: 1e13 1e12 1e11 1e10 MODEL: Dotted z=10 Dashed z=7.5 Solid z=5.5 Dot-dashed z=3.5 DATA: Empty circles z=3 Empty squares z=4 Filled squares z=5 Filled circles z=6 (Mao et al submitted) cfr shapley et al 2001

40. ..and (II) LFs of Ly  emitters MODEL: Dotted z=8 Solid z=6.4 Dashed z=5.7 DATA: Triangles upper and lower limits at z=6.4 circles z=5.7 Additional ingredient here is IGM transmission » 0.5 at z>6.4 and 1 at z=5.7 (Mao et al submitted)

41. Work in progress: Lick spectral indices and colours of Ellipticals (Silva et al. in preparation). Sigma [km/s] Mg1 Red: data Black: models C-M relation

42. CONCLUSION The mutual link between the formation of spheroids and the AGN activity is a key ingredient that must be included into models of galaxy formation. The prescriptions of the ABC scenario (Granato et al. 2001, 2004) lead (in one shot) to predictions in general agreement with many observations which are disturbing for traditional SAMs: statistic of sub-mm galaxies and their mild AGN activity cosmic evolution of QSO activity statistic of massive galaxies at high-z local mass function of SMBH local K band LF of spheroids abundances in ellipticals Main papers: Granato et al. 2001, 2004; Silva et al 2005; Granato et al 2006; Lapi et al 2006, Mao et al 2007, Silva et al in prep Evolution