1. Most star and BH formation is happening in secular evolving disks E. Daddi (CEA Saclay) Mark Sargent(CEA), Mathieu Bethermin(CEA), Giulia Rodighiero(INAF), Georgios Magdis(Oxford), James Mullaney(Durham) GOODS-Herschel team

2. Definition (operational): starburst are the ‘excess SFR’ high-gas-density phase that *can* happen during mergers (or other events) - But does not happen in all mergers - Even in mergers excess phase is short - And in principle might be triggered by other instabilities ? Di Matteo et al 2008 Martig & Bournaud 2008 (Mihos & Hernquist 90’s) Framework: there are 2 ‘major’ SF modes for galaxy buildup: a ‘secular’/’normal’ and a ‘starburst’

3. A ruler:The correlation SFR-M* at different redshifts Z=0, Elbaz et al, 2007 Z=1, Elbaz et al, 2007 Z=2, Daddi et al, 2007 Z=3, Magdis et al, 2010a Z=4, Daddi et al, 2008

4. No main sequence seen if using Herschel data alone (SFR-selection) What is the statistical importance of SB galaxies ? Typical answer: ~50% COSMOS PACS data From PEP (and also GOODS-S) Near-IR galaxies From BzK samples, UV-corrected SFRs Rodighiero et al. 2011

5. Clumpy galaxies are not necessarily mergers, might be extremely rich gas galaxies Bournaud et al 2008 z=1.57 BzK galaxy UDF skywalker See also works of Elmegreen et al., Genzel et al SINS spectroscopy

6. MS outliers: are they mergers ? HGOODS objects with sSFR x4 excess and measured zspec For all cases the UV SFR fails (optically thick) UV underestimate similar to excess sSFR Most likely they are indeed ‘Dense’ mergers  Optically thick sources  Distance from MS

7. PAH distruction in SBsCompactness of SBs Elbaz et al 2011; z=2 Herschel GOODS sample

8. We had lots of IRAM observations to look at multiple CO transitions, and continuum as a byproduct (hard and expensive to get)  This will change dramatically with ALMA Magdis et al 2011 A way to constrain a_CO in the local Universe is through Mdust, As Mdust ~ Z*Mgas Modeling approach: - Full suite of Draine and Lee 2007 Models - MBB with free T and beta

9. Trusting the G/D trend  get estimates for alpha_CO Notice that it is very hard To overestimate much Mgas Hence alpha_CO Because G/D > 1/Z And the local relation has G/D~2/Z Nicely confirming/supporting the bimodality in Mgas/LIR between MS/SB

10. Correlation of LIR/SFR with CO and Mgas (Sargent et al 2012, in preparation) X3 offset from *observables only* X10 offset, dense gas fraction

11. Threshold between MS and SB is objectively defined: 0.6 dex (2.5 sigma of the distribution) Notice that SBs will be present also below the threshold, but becoming overwhelmed by Normal galaxies. Also, they would be objects with minor modification of their SFRs SB: objects with SFR enhanced on average by x4 over what they should heve, given M* Notice very good agreement between UV and Herschel at the high mass bin

12. FIR selection (dust)  SMGs/Herschel galaxies are mixed bags (might explain similarities sometimes found with BzKs e.g. in excitation, SFE)  Need different approach tu build appropriate sample of starbursts SFRD contribution of SBs only ~10% (mergers not so important for star form.) Near-IR selection (stars)  You must be very (un)lucky to pick up a SBs there (2% chance) SB duty cycle ~20Myr Much shorter than ~200Myr typical merger duration (refer only to SFR nnhancement >4)

13. Dec. 5, 201113 (Main sequence + starburst) decomposition 1)Black, white… and grey: bimodality 2)Merger as a Transfer function  MS/SB paradigm using mass functions to (separated)  IR luminosity functions (Sargent+2012) Galaxy counts (Bethermin et al in prep) (SEDs needed – Herschel) H2/CO mass/luminosity functions (Sargent et al) (LCO/H2 to LIR – IRAM) Sargent et al 2012

14. Dec. 5, 201114 IR luminosity function: prediction vs. observations Sargent et al 2012

15. A parameter-less prediction of IR galaxy counts (Bethermin et al 2012, ApJL submitted) Fit is quite good, for fiducial model (no tuning) Not perfect, but much better than earlier attempts already

16. Cosmic evolution of H 2 -reservoirs (Sargent et al 2012 in preparation) Equipartition between molecular gas and stars at z~2

17. How do AGNs fit in all of this ? BH-galaxy correlation (local)  BH assembly must know about galaxy assembly But their BH accretion rate (Lx) doesn’t seem to care about the galaxy (e.g., stellar mass) Mullaney et al 2012

18. Doing ensemble averages (  time averages) All the picture clears up Mullaney et al 2012 astroph Strongly suggest that BH and galaxy do grow Together, and MBH/Mgalaxy is ~constant at the Same ratio of today through formation epoch  There is a Main Sequence also for AGNs! (when eliminating short-time fluctuations) Level is ~ Magorrian

19. Summary and conclusions We have a simple and elegant way to empirically describe the evolution of star forming galaxies through cosmic time, and distinguish the contribution of MS (disk-like, stream-fed?) and SBs (merging driven likely) (2 SF mode framework) We are starting to understand how to distinguish the two modes, at least Globally/statistically, as it is often harder on individual cases. We have testable predictions from this framework, some success (IR LFs, counts, etc) Most of today’s stars were formed in a quiescent mode BHs apparently form in parallel with their hosts, apparently mainly on the MS as well