The 2Df Galaxy Redshift Survey and Cosmological Simulations Carlos S

University of Durham

The 2dF galaxy redshift survey and cosmological simulations Carlos S. Frenk Institute for Computational Cosmology, Durham

Institute for Computational Cosmology The 2dF Galaxy Redshift University of Durham Survey 1997 2002 250 nights at 4m AAT

221,000 redshifts

to bj<19.45

median z = 0.11

First 100k z’s released June/01 Full catalogue released July/03

Institute for Computational Cosmology 2dF Galaxy Redshift Survey: University of Durham Team Members

Ivan K. Baldry10 Carlton M. Baugh2 Joss BlandHawthorn1 Terry Bridges1 Russell Cannon1 Shaun Cole2 Matthew Colless3 Chris Collins13 Warrick Couch5 Nicholas Cross6 Gavin Dalton9 Kathryn Deely5 Roberto De Propris5 Simon P. Driver6 George Efstathiou8 Richard S. Ellis7 Carlos S. Frenk2 Karl Glazebrook10 Edward Hawkins12 Carole Jackson3 Ofer Lahav8 Ian Lewis9 Stuart Lumsden11 Steve Maddox12 Darren Madgwick8 Stephen Moody8 Peder Norberg2 John A. Peacock4 Will Precival4 Bruce A. Peterson3 Mark Seaborne9 Will Sutherland4 Keith Taylor7

Institutions 1AngloAustralian Observatory 2University of Durham 3The Australian National University 4University of Edinburgh 5University of New South Wales 6University of St Andrews 7California Institute of Technology 8University of Cambridge 9University of Oxford 10Johns Hopkins University 33 people at 11University of Leeds 12University of Nottingham 12 13 Liverpool John Moores University Institute for Computational Cosmology institutions The 2dF galaxy redshift survey

University of Durham

QuickTimeª and a YUV420 codec decompressor are needed to see this picture.

Institute for Computational Cosmology The 2dF galaxy redshift survey University of Durham 221414 galaxies ) k ( P

Galaxy power spectrum 3 k

in redshift space, o L convolved with survey 2dF power spectrum 100k window, inclunding non release linear effects Percival etal 01

Institute for ComLputoatgion alk C o(smho loMpgy c1) Galaxy surveys and University of Durham cosmological simulations Simulations are essential for: • Taking into account window function and selection effects • Taking into account nonlinear effects and ‘galaxy bias’ • Assessing systematic and random errors • Comparing data to theory Simulations must: • Be realistic • Have a sufficiently large volume • Resolve the structures of interest ⇒ Large number of particles Institute for Computational Cosmology ΛCDM Hubble Volume Simulation Hubble Volume Nbody simulatn Ω Ω Λ=0.7; m=0.3 σ h=0.7; 8=0.9

9 Np= 10 L = 3000 Mpc

12 mp = 1x10 Mo Virgo consortium (1999)

3000 Mpc/h Real and simulated 2dF galaxy survey

University of Durham

 Real Simulated 

Institute for Computational Cosmology The Millennium simulation University of Durham

Cosmological Nbody simulation • 10 billion particles • 500 h1 Mpc box × 8 1 • mp = 8 10 h Mo UK, Germany, Canada, US Ω Ω Ω collaboration • =1; m=0.25; b=0.045; σ h=0.73; n=1; 8 =0.9

Simulation data available at: × 6 20 10 gals brighter than LMC http://www.mpagarching.mpg.de/Virgo • Carried out at Garching using

Pictures and movies available at: LGadget by V. Springel www.durham.ac.uk/virgo (27 Tbytes of data) Institute for Computational Cosmology QuickTimeª and a 3ivx D4 4.5.1 decompressor are needed to see this picture. The mass power spectrum University of Durham

The nonlinear mass power spectrum is accurately determined by the

Millennium simulation ) “baryon wiggles” k

over large range of ( 2

scales ∆

linear theory

k [h/Mpc]

Institute for Computational Cosmology CMB anistropies and largescale structure University of Durham

2dFgrs

Institute for Computational Cosmology Baryon oscillations in the power spectrum University of Durham

• Predicted by CDM model Ω Ω • Can be used to estimate b/ m • Provide a “standard ruler” that can be used to measure w First tentative detection in 100k 2dFGRS

Institute for Computational Cosmology 100k 2dFGRS power spectrum University of Durham

2dFGRS PS divided by Ωh=0.25 CDM model (zero baryons)

120k redshifts

Percival etal 2001 Institute for Computational Cosmology Baryon oscillations in the power spectrum University of Durham

Oscillations difficult to detect because: • Amplitude is very small • Survey window can smooth over features • Nonlinear effects can erase oscillations • Wiggles are in mass, but we observe galaxies (in zspace)

Need to use large cosmological simulations to establish detectability of baryon oscillations.

Institute for Computational Cosmology University of Durham Baryon oscillations

Baryon oscillations are predicted in the dark matter P(k) in linear theory:

1. Are they erased by nonlinear evolution? 2. Are they also present in the galaxy distribution?

Institute for Computational Cosmology The mass power spectrum University of Durham

Millennium simulation z=127

The Millennium sim is large Simulation k

enough to resolve / )

baryonic wiggles k ( in the matter 2 Linear theory power spectrum ∆

Virgo consortium Springel etal ‘05 k (h/Mpc)Institute for Computational Cosmology The mass power spectrum University of Durham

Millennium simulation z=16 k / ) k ( 2 ∆

Virgo consortium Springel etal ‘05 k (h/Mpc)Institute for Computational Cosmology The mass power spectrum University of Durham

Millennium simulation z=5 k / ) k ( 2 ∆

Virgo consortium Springel etal ‘05 k (h/Mpc)Institute for Computational Cosmology The mass power spectrum University of Durham

Millennium simulation z=0.8

Nonlinear evolution k / accelerates the ) k ( growth of power 2 and eliminates ∆ structure in the spectrum by modecoupling

Virgo consortium Springel etal ‘05 k (h/Mpc)Institute for Computational Cosmology The mass power spectrum University of Durham

Millennium simulation z=0

Nonlinear evolution k / accelerates the ) k ( growth of power 2 and eliminates ∆ structure in the spectrum by modecoupling

Virgo consortium Springel etal ‘05 k (h/Mpc)Institute for Computational Cosmology Baryonic wiggles in the power spectrum University of Durham

Baryonic oscillations in the dark matter PS survive nonlinear effects for log k ≤ 0.7

What about in the galaxy PS?

Institute for Computational Cosmology z = 0 Dark Matter

University of Durham Populating the MS with galaxies

Semianalytic modelling • Find dark matter halos • Construct halo merger trees • Apply SA model (gas cooling, star formation, feedback)

Springel etal 04 Institute for Computational Cosmology z = 0 Galaxy light

University of Durham

Crotton etal 05 Institute for Computational Cosmology University of Durham 14 10 Mo Dark matter Galaxies

Institute for Computational Cosmology Baryon wiggles in the galaxy distribution

Power spectrum from MS divided by a baryonfree ΛCDM spectrum z=7 z=3 Galaxy samples DM gals matched to plausible large observational surveys at given z z=1 z=0 Springel et al 2004 The final 2dFGRS power spectrum University of Durham

2dFGRS 221,000 z’s

Covariance matrix computed from mock catalogues

Cole, Percival, Peacock + 2dFGRS ‘05 Institute for Computational Cosmology The final 2dFGRS power spectrum University of Durham Ω P(k) / Pref( baryon=0) Baryon oscillations conclusively Λ detected in CDM model 2dFGRS!!!

Demonstrates that structure grew by gravitational instability in ΛCDM ΛCDM convolved universe with window

Cole + 2dFGRS ‘05 Institute for Computational Cosmology The final 2dFGRS power spectrum: University of Durham parameter estimation • Shape of P(K) depends on Ωh • Oscillations depend Ω Ω on b/ m • Amplitude depends σ gal on 8

Ωh = 0.161 ± 0.015 Ω Ω ± b/ m = 0.194 0.045

σ gal ± 8 (L*) = 0.870 0.029

Cole + 2dFGRS ‘05 Institute for Computational Cosmology Cosmological parameters: CMB + University of Durham 2dF

Ω Ω Ω σ gal The 2dF power spectrum depends on mh, b / m, 8 , fν, … The CMB power spectrum depends on  , , w , w , f , w ,t , n , n , A ,r , b  k L b dm DE s t s 

Combining 2dF and CMB breaks parameter degeneracies

Institute for Computational Cosmology 2dFGRS + CMB: flatness University of Durham BoomePrraengW, DMASIA,P Maxma, CBI CMB alone has a geometrical degeneracy: large curvature is not ruled out Adding 2dFGRS power spectrum forces flatness: Ω | 1 tot | < 0.04 2dFGRS 100k Efstathiou + 2dFGRS team MNRAS 330, L29 (2002) Institute for Computational Cosmology Spergel etal ‘03 University of Durham

The Guardian Dec/19/2003

Institute for Computational Cosmology University of Durham Cosmological parameters

Pº  , , w , w , f , w ,t , n , n , A ,r ,b  k L b d m DE s t s 

Data:

• CMB → WMAP, CBI, ACBAR and VSA • LSS → Full 2dFGRS

Method:

• MCMC • Use halo Model to relate distribution of galaxies to distribution of mass (galaxy bias) Sanchez, Padilla, Baugh ‘05

Institute for Computational Cosmology Parameter constraints University of Durham Pº  , , w , w ,t , n , A  k L b d m s s CMB only… CMB + 2dF… Λ Λ Ω Ω

Ω Ω k k Sanchez etal ‘05 Institute for Computational Cosmology Parameter constraints University of Durham Pº  , , w , w ,t , n , A  k L b d m s s CMB only… CMB + 2dF… h h

Ω Λ ΩΛ Sanchez etal ‘05 Institute for Computational Cosmology Effect of neutrinos University of Durham

Freestream length: 80 1 (Σ mν /eV) Mpc

Ων=0.05 Ω 2 Σ ( m h = mν / 93.5 eV)

Σmν ~ 1 eV causes lower power at almost all scales, or a bump at the largest scales

100k 2dFGRS Σ  mν < 1.8 ev Eleroy + 2dFGRS team ‘02 Institute for Computational Cosmology Parameter constraints University of Durham Pº  , w , w ,f ,t , n , A  L b dm s s CMB only… CMB + 2dF… ν ν f f

n s ns Σ marginalizing  mν < 1.1 ev (95% c.l.) Institute for Computational Cosmology Parameter estimation University of Durham

CMB + 2dFGRS Ωm = 0.224 ± 0.024 Data: Ωb = 0.055 ± 0.007 → Ω ± • CMB WMAP, CBI, k = 0.034 0.018 ACBAR and VSA ΩΛ = 0.809 ± 0.037 • LSS → 2dFGRS h = 0.683 ± 0.031 Free parameters: ns = 1.07 ± 0.10 Ω Ω Ω 2 Ω 2 τ Λ cdm b k , , h , h , ns, , As σ ± • 8 = 0.812 0.072

Sanchez etal ‘05

Institute for Computational Cosmology Parameter estimation University of Durham

CMB + 2dFGRS For a flat model Data:

Ω ± • CMB → WMAP, CBI, m = 0.238 0.022 ACBAR and VSA Ωb = 0.042 ± 0.003 • LSS → 2dFGRS h = 0.734 ± 0.025 Free parameters: ns = 0.948 ± 0.027

Ω Ω 2 Ω 2 τ • Λ , cdmh , bh , ns, , As

Cole etal ‘05

Institute for Computational Cosmology University of Durham 2dFGRS and galaxy formation

In addition to cosmology, the 2dFGRS contains information about the physics galaxy formation

Institute for Computational Cosmology Efficiency of galaxy formation

University of Durham 15 10 Mo 12 10 Mo 10 10 Mo

Cooling: +++++ Cooling: ++++ Cooling: Feedback: +++++ Feedback: ++ Feedback: ++ Gal. Formation: Gal. Formation: Gal. Formation: ++++ Inefficient Inefficient Efficient Institute for Computational Cosmology Halo masstolight ratios University of Durham E L f

Theoretical / f i M c

i o e prediction l Long τ

a cool n H c y

(semianalytic) of

g Mean a la x y

Galaxy formation f o r is most efficient ma Feedback

12 t i

in ~10 Mo halos o n

Benson, Cole, Baugh, Frenk & 12 1 3 14 1 Lacey 2000, MNRAS H a l o m a s s log M/hInstituMte foor Computational Cosmology Groups and clusters in 2dFGRS University of Durham

Test by:

• Finding groups and clusters in 2dFGRS

• Using simulations to relate groups ←→ dark halos

Institute for Computational Cosmology University of Durham

Mock 2dFGRS groups from Hubble volume N body simulation + semianalytic galaxy formation model

Guest star: Vince Eke

Eke, Frenk, Cole, Baugh +

2dFGRS 2004 Institute for Computational Cosmology University of Durham

Groups in 2dFGRS ≥ 28,213 groups with ngal 2 (53% of gals) ≥ 6,773 groups with ngal 4 ≥ ngal 4: Median z 0.11 Median vel disp 266 km/s Guest star: Vince Eke

Eke, Frenk, Cole, Baugh +

2dFGRS 2003 Institute for Computational Cosmology Halo masstolight ratios University of Durham L E Nmin=2, zmax=0.07 / f M

f i c up i e o r n G c y

Errors in M and L of

cause mocks to g Mean a deviate from la mock x y model prediction f o r M/L overestimated ma t i

10 2 o

for L<10 h Lo n because of scatter in L and errors in M model

Eke etal ‘04 Group luminosity Institute for Computational Cosmology Halo masstolight ratios University of Durham L E / f M

f i c up i e o r n G c y

Errors in M and L of

cause mocks to g Mean a deviate from la mock x y model prediction data f o r ma t i o Mocks and data n agree well!

Eke etal ‘04 Group luminosity Institute for Computational Cosmology Halo masstolight ratios University of Durham

Nmin=2, zmax=0.07 L L E / / f M M

f i Factor of 4 c up up i e o o r r n τ G decrease in M/L G c Long cool y

from rich clusters to of

poor groups g Mean a la mock x y

data f o r Tentative ma Feedback

detection of the t i o minimum n

Eke etal ‘04 Group luminosity Institute for Computational Cosmology How many stars are there

University of Durham and where are there?

Estimate stellar content of 2dFGRS groups using infrared photometry (R from Cosmos; J & K from 2MASS)

Ω ± 3 How many stars? starsh = (0.99 0.03)x10 (Kennicutt IMF)

Galaxy formation theory

⇒ star formation efficiency depends on halo mass

Where are the stars today ? Institute for Computational Cosmology Where are the stars?

University of Durham

l 11 12 13 14

a 10 Mo 10 Mo 10 Mo 5x10 Mo i t

Λ n | | | CDM e Fraction of stars r e

(semianalytic) f in halos of f i

model predicts: d different mass

• Most stellar mass SA model is in LG objects 12 (M~3x10 Mo) e v i t

50% of stellar a • l mass in halos of u 12 m M<5x10 Mo u c • 2% of stellar mass in clusters 14 Log L (group) (M~5x10 Mo) b

Eke etal ‘05 Institute for Computational Cosmology Where are the stars?

University of Durham

l 11 12 13 14

a 10 Mo 10 Mo 10 Mo 5x10 Mo i t

Λ n | | | CDM e Fraction of stars r e

(semianalytic) f in halos of f i

model predicts: d different mass mock • Most stellar mass SA model is in LG objects 12 (M~3x10 Mo) e v i t

50% of stellar a • l mass in halos of u 12 m M<5x10 Mo u c • 2% of stellar mass in clusters 14 Log L (group) (M~5x10 Mo) b

Eke etal ‘05 Institute for Computational Cosmology Where are the stars?

University of Durham

l 11 12 13 14

a 10 Mo 10 Mo 10 Mo 5x10 Mo i t

Λ n | | | CDM e Fraction of stars r e

(semianalytic) f in halos of f i

model predicts: d different mass mock • Most stellar mass SA model is in LG objects 12 (M~3x10 Mo) e v i t

50% of stellar a data • l mass in halos of u 12 m M<5x10 Mo u c • 2% of stellar mass in clusters 14 Log L (group) (M~5x10 Mo) b

Eke etal ‘05 Institute for Computational Cosmology Conclusions: 2dFGRS University of Durham Analysis of 2dFGRS data requires cosmological simulations • Millennium sim ⇒ some baryon oscillations survive in gal distr. From final (221,000 z’s) 2dFGRS :

• Power spectrum ⇒ Consistent with (flat) ΛCDM Significant detection of baryons oscillations • 2dFGRS + CMB ∆T/T ⇒ Ω Ω m=0.238 ± 0.022, b=0.042 ± 0.003, h=0.734 ± 0.0025

⇒ Ω ± 3 • 2dFGRS + 2mass starsh = (0.99 0.03)x10 (Kennicutt IMF) ↑ → • 28,200 2dFGRS ⇒ M/L by x4 from groups clusters groups 12 50% of stars in halos M<5x10 Mo Institute for Computational Cosmology Conclusions: groups and clusters in University of Durham 2dFGRS

• Power spectrum ⇒ Consistent with ΛCDM Significant distortions due to baryons • 2dFGRS + CMB ∆T/T ⇒ Ω Ω Ω m=0.3 ± 0.1, Λ=0.7 ± 0.1, b=0.04 ± 0.01, h=0.70 ± 0.07 evidence for gravitational instability • ξ(σ,π) ⇒ b ≅ 1 ie on large scales gals trace mass

• Semianalytic model ⇒ Clustering of halos Powerlaw ξ(r): a coincidence Occupation stats P(N,M)

Lum fn of all galactic systems Agree with 2dFGRS groups ΛCDM (SA) • M/L as a fn of Mhalo

P(N,M) Institute for Computational Cosmology