Published on August 29, 2007
High redshift (z~4) galaxies& clustering: High redshift (z~4) galaxies andamp; clustering Lexi Moustakas STScI credit: credit Everybody at GOODS andamp; ODT! Soo Lee (JHU) (advisor: M. Giavalisco) Paul Allen (MSO, [email protected]) Emily MacDonald (Oxf) (advisor: G. Dalton) see M. Giavalisco talk tomorrow!: GOODS: Giavalisco et al 2004 Montage courtesy of F. Summers total GOODS: ~320 arcmin2 see M. Giavalisco talk tomorrow! Finding high-z galaxies: z~4: Finding high-z galaxies: z~4 The Lyman-dropout technique, B-V vs V-z (for z~4) -- multiwavelength is KEY The space-based GOODS data use the z-band andamp; are extremely deep compared to the ground -- ~2-3 mag fainter. In total GOODS ACS area, ~2000 z~4 galaxies B-dropouts, z~4 Giavalisco et al. 2004 LBG redshift distributions,from monte carlo simulations: LBG redshift distributions, from monte carlo simulations The redshift distributions are well-constrained through simulations. The completeness is more difficult to pin down. (The B-drops are the z~4). Giavalisco andamp; S. Lee 2004 morphologies of faint z~4 galaxies: morphologies of faint z~4 galaxies The sizes of star forming galaxies above z~1 are sub-arcsec (Ferguson et al 2004) As shown here, the morphologies are varied and can be complex The pair/group statistics are crucial for characterizing environment Viz 1'' from the v1.0 GOODS data Check out the scale! clustering of faint z~4 galaxies: clustering of faint z~4 galaxies With the angular correlation function measured directly, and a simulated N(z), we invert andamp; calculate the spatial correlation function x(r) = (r/r0)-g , usually assumed to be a power-law on relatively large scales, with characteristic scale r0. S. Lee et al. 2004, in prep. w(theta) vs angular separation } nb: many neighbors within 10-20arcsec! clustering with app. magnitude: clustering with app. magnitude Clustering measured in the GOODS data to different magnitude limits. (The error bars are smaller than the points!) There is evidence for stronger clustering in the brighter samples... (See also Giavalisco andamp; Dickinson 2001). GOODS data from S. Lee et al. 2004, in prep. spatial clustering vs limiting apparent magnitude clustering with abs. magnitude: clustering with abs. magnitude Transform (approximately) to rest-frame BJ magnitudes The brightest point is sub-L* What happens if one goes to much brighter absolute magnitudes?? =andgt; We don't know from GOODS! Area is not large enough to find very rare objects... spatial clustering vs absolute magnitude (approximate) The Oxford-Dartmouth Thirty-Degree (ODT) SurveyMacDonald et al 2004, MNRAS, in press: The Oxford-Dartmouth Thirty-Degree (ODT) Survey MacDonald et al 2004, MNRAS, in press 5s limits completion vega to date U andgt; 25 B 26.0 V 25.5 R 25.25 andgt;23 deg2 i 24.5 Z 22 K andlt; 19 andgt; 3.5 deg2 MacDonald et al. 2004 Moustakas et al in prep (K-band part) andr 0018+3452 lynx 0909+4050 herc 1639+4524 virgo 1200+0300 The ODT Survey: A wide-field multi-l survey: The ODT Survey: A wide-field multi-l survey The Andromeda field of the ODT Survey A GOODS Field MacDonald et al. 2004 clustering of bright z~4 galaxies: clustering of bright z~4 galaxies Clustering measurements of B-drops in ODT Survey, from a ~2deg2 subset Allen et al. 2004, MNRAS N(z)'s 'realized', and angular correlation function inverted. These LBG samples are bright, with iandlt;24.5 (2mag brighter than GOODS) Allen et al. 2004 } nb: no neighbors within 10-20arcsec! L-dependent clustering at z~4: L-dependent clustering at z~4 GOODS: S. Lee et al. ODT: P. Allen et al. L-dependent clustering at z~4: L-dependent clustering at z~4 L* is around here GOODS: S. Lee et al. ODT: P. Allen et al. L-dependent clustering at z~0: L-dependent clustering at z~0 z~0 GOODS: S. Lee et al. ODT: P. Allen et al. 2dF: Norberg et al. 2002 cosmic variance in this result: cosmic variance in this result Assuming simple galaxy-halo correlation larger volumes = less cosmic variance smaller clustering = less cosmic variance We calculate a similar level of cosmic variance across the z~4 result -- GOODS: small volume but small clustering -andgt; cv~20% ODT-S: large volume but large clustering -andgt; cv~40% To bring the high-L variance down to 20%, need andgt;10 times more area! But even that isn't enough. Why is that? -- Onwards, to: beyond sweet peas: beyond sweet peas Clustering, (dark matter) masses, and environment With analytic LCDM, we can connect the clustering to the minimum dark matter halo mass. Combining the clustering with the space densities, a Halo Occupation Distribution (HOD) formalism can constrain the number of galaxies per halo vs halo-mass Adding luminosity information to this, the Conditional Luminosity Function (CLF) Let's quickly consider the Halo Occupation formalism dark matter halo masses: dark matter halo masses Moustakas andamp; Somerville 2002 There can be many galaxies in each dark matter 'halo', or none. The average behavior can be parametrized with the Halo Occupation Function, or Distribution N(Mandgt;Mmin) = (M/M1)a Mmin - threshold halo mass ** from clustering M1 - 'typical' mass ** from clustering andamp; density a - mass function slope ** from small-scale clustering! 'bias' comes from the clustering, which fixes the 'minimum' DM halo mass space density bias galaxies' dark matter halos: galaxies' dark matter halos Here we plot the results for z~0 ellipticals, z~1.2 EROs, and z~3 LBGs (LAM andamp; Somerville '02) The occupation function parameters can be constrained through the measured clustering strength and the space density The SLOPE (a 'free' parameter in this plot), can be constrained by very small-scale statistics Mandamp;S02 clustering evolution: clustering evolution The simplest model hasa galaxies following the dark matter they're associated with -- 'galaxy conserving model' (Fry 1996) See the behavior of populations with properties established at different redshifts. Do they 'connect'? correlation scale linear bias Conclusions: Conclusions There is evidence for luminosity- dependent clustering in galaxies, at z~4 as well as locally Need 'complete' census at all scales =andgt; DEPTH andgt;10s of square degrees or more will be required to characterize this: =andgt; LARGE SOLID ANGLE To constrain the SLOPE of the occupation function, we need very sub-few-arcsec pair/group info.: =andgt; HIGH SPATIAL RESOLUTION A multi-wavelength SNAP/JDEM/LEGASY type mission would clean this up...