massive galaxies massive galaxies at high redshift:at high redshift:models confront observationsmodels confront observations

rachel somervilleSTScI

rachel somervilleSTScI

GIRLS; Leiden, September 2008

z=5.7 (t=1.0 Gyr)z=5.7 (t=1.0 Gyr)

z=1.4 (t=4.7 Gyr)z=1.4 (t=4.7 Gyr)

z=0 (t=13.6 Gyr)z=0 (t=13.6 Gyr)

Springel et al. 2006Springel et al. 2006 Wechsler et al. 2002

shock heating & radiative cooling

photoionization squelching

merging star formation (quiescent & burst)

SN heating & SN-driven winds

chemical evolution

stellar populations & dust

shock heating & radiative cooling

photoionization squelching

merging star formation (quiescent & burst)

SN heating & SN-driven winds

chemical evolution

stellar populations & dust

rss, Hopkins, Cox, Robertson & Hernquist 2008, MN in press

fairly broad consensus: SN-driven winds remove baryons in small-mass halos

some process(es) prevent(s) cooling in large-mass halos (radio jets, clumps, conduction, cosmic ray pressure?)

fairly broad consensus: SN-driven winds remove baryons in small-mass halos

some process(es) prevent(s) cooling in large-mass halos (radio jets, clumps, conduction, cosmic ray pressure?)

quenching of massive galaxies

(note the slope is wrong for low mass galaxies.this is not due to AGN FB, & cannot be easily solved by ‘tweaking’)

rss, Hopkins, Cox, Robertson & Hernquist 2008, MN in press

SS

FR

stellar mass

halos with primarily “cold” vs. “hot” flows separated by a critical mass of few x 1012 Msun at low redshift (e.g. Birnboim & Dekel 2003; Keres et al. 2004);

heating processes only effective when a quasi-static hot gas halo is present (i.e. in large mass halos)

halos with primarily “cold” vs. “hot” flows separated by a critical mass of few x 1012 Msun at low redshift (e.g. Birnboim & Dekel 2003; Keres et al. 2004);

heating processes only effective when a quasi-static hot gas halo is present (i.e. in large mass halos)

hot vs. cold flowshot vs. cold flows

simulations: A. Kravtsov

M*

Mvir [Mʘ]

all hot

1014

1013

1012

1011

1010

109

0 1 2 3 4 5redshift

z

all cold

cold filamentsin hot medium

MshockMshock>>M*

Mshock~M*

Dekel,Birnboim,Zinger, Kravtsov

dense cold filaments can penetrate the hot mediumin large-mass halos at high redshift

associated with optical/X-ray luminous AGN/QSO

triggered/fed by mergers or secular (bar) instabilities

high accretion rates (0.1-1 LEdd), fueled by cold gas via thin accretion disk

may drive winds that can shut off further accretion onto the BH and sweep the ISM out of the galaxy

associated with optical/X-ray luminous AGN/QSO

triggered/fed by mergers or secular (bar) instabilities

high accretion rates (0.1-1 LEdd), fueled by cold gas via thin accretion disk

may drive winds that can shut off further accretion onto the BH and sweep the ISM out of the galaxy

QSO/bright mode

Radio Mode radio galaxies

(classical FR I and FR II type sources); generally no optical emission lines

‘low accretion state’ (low Eddington ratio, <10-3 Bondi accretion or ADAF?)

jets may heat gas in a hydrostatic hot halo, offsetting or quenching cooling flow

radio galaxies (classical FR I and FR II type sources); generally no optical emission lines

‘low accretion state’ (low Eddington ratio, <10-3 Bondi accretion or ADAF?)

jets may heat gas in a hydrostatic hot halo, offsetting or quenching cooling flow

star formation history stellar mass build-up

all stars

star formationin bursts

blue: fiducial model(cLCDM 8=0.9)red: WMAP3orange: no cooling if Mh<1011 Msun

time-dependent IMF?SFR at high-z overestimated?

solid: MORGANAdash: Munich Mill.dot-dash: rss08

stellar mass function evolution

“raw” model predictions with convolved errors

Fontanot, de Lucia, Monaco & rss in prep

stellar mass assemblywithout mass errors with errors (0.25 dex)

solid: MORGANAdash: Munich Mill.dot-dash: rss08

data: red: Conselice et al.blue: composite MF

Fontanot et al. in prep

star formation rate density as function of galaxy mass

green: GOODS; blue: Zheng et al. (COMBO-17)red: Conselice et al.; cyan: Mobasher et al. 2008

solid: MORGANAdash: Munich Mill.dot-dash: rss08

Fontanot et al. in prep

data:red square: Drory et al. 2008blue: Bell et al. 2007cyan: Martin et al. 2007green: Grazian et al. 2006magenta: Noeske et al. 2007red x: Chen et al. 2008blue diamond: Dunne et al. 2008

evolution of the SF ‘main sequence’

Fontanot et al. in prep

archeological downsizingarcheological downsizingdata: Panter et al. 2007 data: Gallazzi et al. 2007

Fontanot et al. in prep

when did the red sequence emerge?when did the red sequence emerge?

the red sequence is still clearly identifiable in the field & clusters up to z~1 (Bell et al. 2005; Faber et al. 2007)

recently, a population of massive red galaxies detected in the field at 2<z<3 (Kriek et al. 2008; Taylor et al. 2008)

very red populations discovered in clusters up to z~2, but absent by z~3 (Zirm et al. 2008; Kodama et al. 2008)

the red sequence is still clearly identifiable in the field & clusters up to z~1 (Bell et al. 2005; Faber et al. 2007)

recently, a population of massive red galaxies detected in the field at 2<z<3 (Kriek et al. 2008; Taylor et al. 2008)

very red populations discovered in clusters up to z~2, but absent by z~3 (Zirm et al. 2008; Kodama et al. 2008)

Kriek et al. 2008

U-B

Zirm et al. 2008

fieldpopulation

color-magnitude relation for comacolor-magnitude

relation for coma

Trager & rss 2008

black points: SDSSred points: SAM

rest-frame u-r for proto-clusters (M>1014 Msun)

‘field’ RSfrom Tayloret al. ECDFS

z~0 SDSS RS

Millenniumz=2 clusters

Zirm et al.data

H(AB)

observed frame J-H for proto-clusters (M>1014 Msun)

z=2

Testing physical parameter extraction from broad-band

photometry

Testing physical parameter extraction from broad-band

photometry

created SAM mock catalogs (including dust & IGM) and extracted U-, B-, and V-dropouts using GOODS selection criteria

added photometric errors by bootstrapping from GOODS data

ran a fairly standard BC03-based SED-fitting code on ACS+ISAAC+IRAC photometry (extract stellar mass, stellar population age, and SFR)

created SAM mock catalogs (including dust & IGM) and extracted U-, B-, and V-dropouts using GOODS selection criteria

added photometric errors by bootstrapping from GOODS data

ran a fairly standard BC03-based SED-fitting code on ACS+ISAAC+IRAC photometry (extract stellar mass, stellar population age, and SFR)

S. Lee, R. Idzi, H. Ferguson, rss, T. Wikland, M. Giavalisco

U-drops (z~3); redshift fixed

-19%

-65%

x2

B-drops (z~4); redshift fixed

-25%

-58%

x2

B-drops; redshift fit

U-drops with largest mass errors U-drops with smallest mass errors

Stringer et al. 2008

z~0.4-1.4DEEP+Palomar photometryfixed redshiftbootstrapped photometric errorsBundy et al. (2006) massestimation method

no significant offset or mass trend--> scatter ~0.15 dex

parameter estimation summary

parameter estimation summary

sources of error: mismatch between assumed “tau” SFHs and SAM predicted SFH

‘hiding’ of mass beneath young stellar population

‘conspiracy’ of overestimated age (--> higher mass estimate) + ‘youth bias’ (lowers mass estimate) actually reduces mass errors

two-component models (with ‘maximally old’ component or secondary burst) produce improved age & SFR estimates, but poorer mass estimates!

sources of error: mismatch between assumed “tau” SFHs and SAM predicted SFH

‘hiding’ of mass beneath young stellar population

‘conspiracy’ of overestimated age (--> higher mass estimate) + ‘youth bias’ (lowers mass estimate) actually reduces mass errors

two-component models (with ‘maximally old’ component or secondary burst) produce improved age & SFR estimates, but poorer mass estimates!

summarysummary differences between observational datasets much larger than differences between models!

number/mass density of massive galaxies is reproduced fairly well by models (when mass errors convolved) to z~2

SFR of massive galaxies at z~1-2 underestimated by factor of ~few in models if observational estimates taken at face value (IMF, AGN contamination, large errors in SED-fit based estimates?)

low mass galaxies form too early in models --> mass assembly “upsizes” rather than “downsizes”

massive galaxies in large mass halos are being quenched too late in models (RS emerges late)

errors in stellar masses, SFR, and ages derived from SED fitting to broad-band photometry at high redshift may be larger than we think...

differences between observational datasets much larger than differences between models!

number/mass density of massive galaxies is reproduced fairly well by models (when mass errors convolved) to z~2

SFR of massive galaxies at z~1-2 underestimated by factor of ~few in models if observational estimates taken at face value (IMF, AGN contamination, large errors in SED-fit based estimates?)

low mass galaxies form too early in models --> mass assembly “upsizes” rather than “downsizes”

massive galaxies in large mass halos are being quenched too late in models (RS emerges late)

errors in stellar masses, SFR, and ages derived from SED fitting to broad-band photometry at high redshift may be larger than we think...

bias in line-strength derived ages

bias in line-strength derived ages

stellar mass

mass weighted age

LS derived age

for 20 realizations ofa Coma-sized halo Trager & rss 2008

star formation histories of early type galaxiesas a function of stellar mass

SF histories of E’s in hierarchical models show qualitatively correct ‘downsizing’ behavior

but, probably not strong enough (new evidence from /Fe ratios -- Arrigoni, Trager & rss in prep)

SF histories of E’s in hierarchical models show qualitatively correct ‘downsizing’ behavior

but, probably not strong enough (new evidence from /Fe ratios -- Arrigoni, Trager & rss in prep)

the original downsizing plot

the original downsizing plot

Cowie et al. 1996

~SSFR

~stellar mass

the many manifestations of

‘downsizing’

the many manifestations of

‘downsizing’ SF history from lookback studies (original Cowie definition): star formation activity shifts to lower mass galaxies over time

mass assembly histories: high mass galaxies assembled early, low mass galaxies assembled later

archeological downsizing: stellar ages are younger in low mass galaxies, indicating a later epoch of SF

chemo-archeological downsizing: higher [/Fe] ratios in more massive galaxies indicate a shorter epoch of formation

SF history from lookback studies (original Cowie definition): star formation activity shifts to lower mass galaxies over time

mass assembly histories: high mass galaxies assembled early, low mass galaxies assembled later

archeological downsizing: stellar ages are younger in low mass galaxies, indicating a later epoch of SF

chemo-archeological downsizing: higher [/Fe] ratios in more massive galaxies indicate a shorter epoch of formation