From Fine Structure to LargeScale Structure with the DEEP2 Redshift Survey

1
From Fine Structure to Large-Scale Structure with
the DEEP2 Redshift Survey
Jeffrey Newman Lawrence Berkeley National
Laboratory
2
The DEEP2 Collaboration
The DEEP2 ( DEEP Extragalactic Evolutionary
Probe 2) Galaxy Redshift Survey is studying both
galaxy properties and large-scale structure at
z1.

• U.C. Berkeley
• M. Davis (PI)
• D. Croton
• M. Cooper
• B. Gerke
• R. Yan
• C. Conroy
• LBNL
• J. Newman

U.C. Santa Cruz S. Faber (Co-PI) D. Koo P.
Guhathakurta D. Phillips K. Noeske A. Metevier
L. Lin N. Konidaris G. Graves

• Steward Obs.
• A. Coil
• C. Willmer
• Princeton
• D. Finkbeiner
• Maryland
• B. Weiner
• U. Pitt.
• A. Connolly

3
Outline

• I. The DEEP2 Redshift Survey
• Testing for variations in the Fine Structure
  Constant
• Using large-scale structure in DEEP2 to study the
  evolution of red galaxies
• Further insights from SDSS

4
Redshift surveys in brief

• Redshift surveys map out the large-scale
  structure traced by galaxies (or QSOs) by
  locating them in 3d space. Over the past 25
  years, surveys of the local Universe have
  progressed from 2500 galaxies in a thin slice of
  sky (Davis et al.s CfA1) to 106 over 1/4 of
  the sky (Sloan Digital Sky Survey)

2005
1982
5
Surveying distant galaxies can constrain both
cosmology and galaxy evolution
The evolution of the pattern of filaments and
voids traced out by galaxies - the large-scale
structure - is strongly dependent on the
underlying cosmology. By comparing the universe
at high redshift to z0, we can both perform
cosmological tests and simultaneously study
galaxy formation and evolution.
6
Scientific Goals of the DEEP2 Galaxy Redshift
Survey

• Characterize the properties of galaxies (colors,
  sizes, linewidths, luminosities, etc.) at z1 for
  comparison to z0
• Study the clustering statistics (2- and 3-pt.
  correlations) of galaxies as a function of their
  properties, illuminating the nature of the galaxy
  bias
• Determine N(s,z) of groups and clusters at high
  redshift, providing constraints on ?m and w
• Measure the small-scale thermal motions of
  galaxies at z1, providing a mass scale for halo
  models (measuring ?m and bias, in the paradigm
  when DEEP2 was designed)

7
Vital statistics of DEEP2

• 3 square degrees of sky
• 4 fields (0.5o x lt2o)
• 400 slitmasks observed over 80 Keck nights
• 1200 l/mm grating 6500-9200 Å
• 1.0 slit FWHM? 68 km/s
• Primary redshift range0.7-1.4 (pre-selected
  using BRI photometry)
• gt40,000 redshifts
• 5106 h-3 Mpc3 comoving volume
• One-hour exposures
• RAB24.1 limiting magnitude

8
Comparison to Local Surveys
DEEP2 was designed to have comparable size and
density to previous generation local redshift
surveys and is gt50 times larger than previous
surveys at z0.3-1.
SDSS
DEEP2 is between LCRS and 2dF in sample size, but
at z1
2dF
LCRS
DEEP2
z0 z1
CFA SSRS
PSCZ
9
DEEP2 has been made possible by DEIMOS, a new
instrument on Keck II
DEIMOS (PI Faber) and Keck provide a unique
combination of wide-field multiplexing (up to 160
slitlets over a 16x4 field), high resolution
(R5000), spectral range (2600 Å at highest
resolution), and collecting area.
10
DEEP2 slitmask spectroscopy
l
position
A total of 400 slitmasks are required for DEEP2
we can tilt slits up to 30 degrees to obtain
rotation curves.
11
Pre-selection of high-z targets with using colors

• Plotted are the colors of galaxies with known
  redshifts in our fields those at low redshift
  are plotted as blue, those at high redshift as
  red. We use a simple color cut defined by three
  line segments to select galaxies at zgt0.75.
• We do not apply these color cuts in the EGS!

12
Redshift Distribution of Data z0.7-1.4
90 of our targets are at zgt0.75 and we miss
only 3 of high-z objects.

• Status
• designed as a
• three-year survey
• - began summer 2002
• - currently gt95 complete
• finished 3 of 4 fields, 4th
• will be completed spring 2006

13
AEGIS the All-wavelength Extended Groth Strip
International Survey
Spitzer MIPS, IRAC
Background 2 x 2 deg from POSS
DEEP2 spectra and Caltech / JPL Ks imaging
HST/ACS V,I (Cycle 13)
Plus VLA (6 21 cm), SCUBA, etc.
14
By combining area with depth, AEGIS allows us to
study rare objects in detail
Gerke et al. 2006, in prep.
15
Like a spectroscopically identified, dual AGN
Hb
OIII 4959
OIII 5007
position
l/ z
16
We have measured its SED over 9 decades in n
17
And have high-resolution HST imaging
18
Some useful numbers

• Age of the Universe 13.7 Gyr
• z0.7 6.3 Gyr ago
• z1.0 7.7 Gyr ago
• z1.4 9.1 Gyr ago
• Within DEEP2 we are surveying
• gt2.5 Gyr or 20 of the history
• of the Universe. SDSS/2dF comparisons give 3x
  this baseline.

19
Are the fundamental constants of Nature the same
everywhere in the Universe and over all times?
DEEP2 data can be used to answer questions not
considered when the survey was designed. For
instance, we have now used DEEP2 to test for
variation in the Fine Structure Constant, ?.
20
A Quick Review
The fine-structure constant is
(cgs units) It
is the dimensionless coupling constant of
QED--i.e., it determines the strength of
electromagnetism. Its measured value at the
present day is 1/137.03599911.
21
Why test for changes in a?

• ? provides some of the simplest tests of the
  universality of physical laws.
• Temporal and/or spatial variation in ? is
  predicted by some dark energy scenarios and
  theories with large extra dimensions.
• There have been recent claims that significant
  evolution has been detected.
• Most methods of testing for evolution in ? are
  likely dominated by unknown systematics
  (e.g.,different groups get contradictory results
  with same basic method).

22
In memory of a pioneer

• Shortly after quasars were first identified, the
  late John Bahcall used them to test for evolution
  in a using the wavelengths of two types of line
  doublets
• Oxygen III emission from the quasar
• Metal (Fe,Si,etc.) absorption from foreground
  gas clouds
• All astrophysical tests for varying a come from
  these roots. For the last 30 years,
  absorption-line methods have dominated the field.

23
Recent measurements disagree with each other
Oxygen emission lines provide lower nominal
precision than absorption line/many-multiplet or
natural reactor methods, but the (astro)physics
is much simpler.
24
Physics of OIII and a
To lt1, a2 ? (l2-l1)/(l2l1) for OIII 4959
5007 Å emission lines, as their splitting arises
from fine structure directly.
As they are forbidden (so extremely optically
thin) and emitted from the same state, these
OIII emission lines must have line profiles
proportional to each other (regardless of gas
density or velocity, etc).
25
Effect of changing a
Shown is the effect of a 5 change in a applied
to an actual spectrum - we can detect evolution
1/800 this large!
26
Testing ? with DEEP2 vs. SDSS
The DEEP2 sample is unique in using galaxies,
rather than the brighter but rarer QSOs, to study
a. Advantages of DEEP2 over SDSS for this work
include

• Higher spectral resolution (0.3 Å/pix).
• Attendant tighter control of wavelength
  solution. Systematics are lt0.001Å differential
  between the two OIII lines.
• Larger sample (858 galaxies vs. 23-308 QSOs),
  allowing better tests of errors.
• Higher typical redshift (median z0.72 vs 0.37
  OIII is present in DEEP2 spectra for
  0.28ltzlt0.80).

Main disadvantage is lower S/N for each
individual object - so we beat everything down by
vN.
27
We must control systematic wavelength calibration
errors
Temperature fluctuations, focus changes, etc.,
can alter the relationship between pixel
wavelength from when we calibrated it with
arclamps in the afternoon. We remedy this by
cross-correlating the sky spectrum from each slit
vs. a high-quality template and solve for shift
vs. wavelength.
28
Combining info from many slits helps
RMS residual of single-slit fits about the global
solution for all slits is 0.006-0.008Å -
dominated by the individual measurement
uncertainties. Actual errors should be 1/10 as
large (as gt100 slits in fit).
Residuals of individual measurements from full
1d/2d correction
N
Initial systematic offsets
Corrected wavelength errors are dwarfed by
centroid errors--were limited by S/N of emission
lines.
Difference in l (Å)
29
We detect no change in a from z0 to z0.7
Start with the simplest thing combine all
galaxies with zgt0.6 into one bin, and measure
ltDgtltDa2/a2gt.
Newman et al. 2006, in prep
We measure Da/a -8.08 ?10-6 /- 1.91 ?10-5 at
median z0.72
30
Results - checking for an offset
Results are consistent with no evolution in a
from z0. We have tried to perturb many of the
ways we do things and this remains true.
31
Changing fraction of sample used
32
We also detect no significant slope, da/dt
We measure da/dt 9.55?10-15 /- 2.58 ?10-14 yr-1
33
Results da/dt
Null hypothesis (no evolution from z0) has
reasonable c2 . Adding 1 or 2 parameters only
improves c2 marginally.
34
DEEP2 vs. previous measurements
Nominal precision does not approach QSO
absorption-line measurements. However, OIII
method is much simpler and should be more robust.
Future surveys could expand samples greatly!
35
Testing for spatial variation
We can test for spatial variation in ? by
measuring the differences in ltDa2/a2gt amongst
similar volumes in the 4 DEEP2 fields at the same
redshifts.
10 h-1 Mpc
We set a 95 upper limit on Gaussian spatial
fluctuations at 1000-3000 Mpc separations s
(Da/a) lt9.0?10-5
36
We detect no spatial variation, either
We set a 95 upper limit on Gaussian spatial
fluctuations at 1000-3000 Mpc separations
s(Da/a) lt9.0?10-5
37
The Future
Number of OIII-emitting galaxies needed to rule
out (or confirm) the many-multiplet detections at
99 confidence
Cf. 540 objects (1 of full sample) in DEEP2
measurement. However,
future surveys will likely target bright
emission-line galaxies over a narrower redshift
range - just what is needed!
38
The origins of red galaxies

• Although classically, galaxies were divided into
  classes by morphology (e.g. spiral/elliptical),
  we can more robustly measure their colors,
  especially at high z.
• All galaxies which have not formed new stars for
  ? 1Gyr have very similar spectra/colors, as both
  the surviving low-mass stars and red giants which
  dominate their emission have similar T.

This red galaxy population corresponds closely to
the classical early type galaxies elliptical,
S0s, etc. They include the most massive galaxies
known.
39
Why are red galaxies interesting?

• The red population appears much simpler than the
  heterogeneous blue, star-forming galaxies - they
  can be represented well by a 1- or 2- parameter
  family.
• However, models of galaxy formation and
  evolution have difficulty producing truly
  quiescent, non-starforming red galaxies,
  particularly in the numbers seen.

40
A well-defined red sequence exists to z1.4
brighter
Bimodality in color persists to the limits of the
survey, and is a vital tool for many analyses.
Note Our R-band limit corresponds to 4000Å
rest-frame at z0.7, 2800 Å at z1.4, so we lose
red galaxies before blue ones.
Willmer et al. 2005
41
Where do typical RS / early-type galaxies come
from?
The total comoving abundance of red galaxies
changes by 4? between z1 and z0, while the
abundance of blue galaxies changes little over
this time. The red population is not just
evolving passively from z3!
log ? ()
z
Faber et al. 2005
42
3 basic scenarios for quenching red galaxies

• Major mergers trigger bursts of
  star-formation/SNe or luminous AGN (QSOs) that
  blow out gas over a large volume (e.g. Hopkins et
  al. 2005), removing the fuel for SF.

43
3 basic scenarios for quenching red galaxies

• Mechanisms which are efficient in massive
  clusters (cooling suppression by the cluster
  medium, ram pressure stripping, galaxy
  harassment) halt star formation after galaxies
  fall in.

44
3 basic scenarios for quenching red galaxies
Keres et al. 2005

• In halos with masses above 3?1011 Msun, gas
  cannot cool efficiently (Silk 1977, White Frenk
  1991, Keres et al. 2005, etc.), so star formation
  can be quenched.

45
An extra ingredient ?
In this last scenario, a low-level (10-5
Eddington) AGN may still be needed to keep any
star formation from occurring at late times in
massive galaxies (e.g. Croton et al. 2005).
46
Placing RS galaxies in LSS context can help
Structure seen in DEEP2 7 Gyr ago is similar to
that in SDSS. We can now study how the evolution
of galaxies relates to large-scale structure.
47
Galaxy Properties and Environment

• We can measure the local density - i.e., the
  environment of any given object - using
  galaxies with redshifts as a tracer of matter.

log overdensity
linear overdensity
blue color red
blue color red
Cooper et al. 2006, submitted
48
Environment over the CMD
SDSS, z0.1
DEEP2, 0.75ltzlt1.05
redder
brighter

• Basic trends from z0 studies persist at z1
  e.g., the reddest and brightest galaxies are
  preferentially found in dense environments.

Cooper et al. 2006, submitted
49
Environment vs. Luminosity
Blue galaxies
Red galaxies
denser
brighter
brighter

• However, unlike locally, red and blue galaxies
  have very similar trends of environment vs.
  luminosity at z1.

50
In other words
denser
brighter

• There exists a population of bright, blue
  galaxies in dense regions that is present at z1
  but not today. Presumably, they have quenched
  and are now on the red sequence.

51
Baby pictures of todays RS galaxies?
52
Galaxy groups in DEEP2

• We can focus on galaxy populations in the densest
  regions by identifying groups of galaxies. We use
  overdensities in the galaxy distribution in
  redshift space to locate them, using the VDM
  algorithm of Marinoni et al. (2002).

Group in early DEEP2 data ?250 km/sec
53
Why look for groups clusters?
We are identifying groups in DEEP2 not only to
study galaxy evolution, but also because their
apparent abundance provides a test of dark energy
models. See Newman et al. 2002 and Gerke et al.
2005 for details.
54
Color/environment trend is driven by group (not
massive cluster) galaxies
Mean median trends
log overdensity
After group and cluster galaxies are removed
blue red
Cooper et al. 2006, submitted
55
Field blue-galaxy luminosity trend resembles
hierarchical picture
Mean trend
log overdensity
After group and cluster galaxies are removed
brighter
Cooper et al. 2006, submitted
56
Do these trends evolve over time?

• Sample definition is critical for a clean test of
  the Butcher-Oemler effect.

brighter
Gerke et al. 2006, in prep.
57
Evolution of blue fraction in groups
The blue fraction is lower in groups than the
field, but evolves more quickly, and appears to
be converging at z1.2. This would suggest that
galaxies in groups start quenching at z1.5 or so.
z
Gerke et al. 2006, in prep.
58
Can we find galaxies quenching onto the red
sequence?
Some star formation indicators (e.g Ha) go away
quickly (100 Myr), while A stars last 1 Gyr.
We can use this to find galaxies that have
recently ceased star formation post-starburst
or KA galaxies.
HST images of DEEP2 KAs
59
At zgt0.3, OII is used to define KAs
but in SDSS, post-starburst galaxies stand out in
H? but not in OII.
60
Curiouser and curiouser
OII emission in post-starburst galaxies can be
as strong as in star-forming galaxies.
Yan et al. 2006, submitted, astro-ph/0512446
61
OII/H? Bimodality
The full SDSS sample displays a bimodality, too.
Yan et al. 2006,astro-ph/0512446
62
Color Bimodality
Red Galaxies
Blue Galaxies
brighter
63
OII/H? bimodality echoes color bimodality
Yan et al. 2006,astro-ph/0512446
64
Color-Magnitude Distributions divided up by line
ratio
brighter
Yan et al. 2006,astro-ph/0512446
65
High-ratio galaxies are LINERS or quiescent
High-OII/H?
Red-Low-OII/H?
Yan et al. 2006,astro-ph/0512446
66
Environment provides a clue to the LINER
mechanism!
Yan et al. 2006
Blanton et al. 2004
brighter
The brightest red galaxies are quiescentand in
densest regions!
67
Conclusions

• We find no temporal or spatial variations in a.
• Trends between properties and environment for
  typical galaxies were in place by z1. They are
  driven by group-scale, not cluster-scale,
  phenomena.
• Groups appear to have become a suitable
  environment for quenching around z1.5.
• LINERs are very frequently found in red galaxies.
  
• Our results are broadly consistent with models
  where both group-mass halos and AGN are required
  for keeping typical red galaxies quenched.
• Survey data can have wide applications, far
  beyond what we originally planned!

68
Possible systematics

• To change Da2 by 0.5s, a systematic must change
  (l2-l1) by 0.0015Å rest frame, or (l2l1)/2 by
  0.15Å rest frame. Systematics tested so far are

69
OIII vs. Many-multiplet
Bahcall et al. 2003
The first attempts to look for variations in a
with QSOs used OIII emission (e.g. Bahcall
Salpeter 1965). Absorption lines became
preferred as they could yield more measurements
per spectrum (limited by of absorbers, while
each object has 1 redshift) and can be observed
to higher z.
70
Local LINERs have the right luminosities to be
quenching AGN
Ho 2005
71
Other recent and upcoming papers include

• Angular clustering of galaxies Coil et al.,
  2004, ApJ, 617, 765
• DEEP2 survey strategy dark energy Davis et
  al.,astro-ph/0408344
• Evolution of close-pairs/merger rates Lin et
  al., 2004, ApJ, 617, 9
• DEEP2 Group catalog Gerke et al., 2005, ApJ,
  625, 6
• Satellite galaxy kinematics Conroy et al., acc.,
  astro-ph/0409305
• Environment in deep redshift surveys Cooper et
  al., acc. (0506518)
• DETF white paper Davis et al., astro-ph/0507555
• Void statistics Conroy et al., acc.,
  astro-ph/0508250
• Luminosity function Willmer et al., acc. Faber
  et al., submitted
• Metallicities of DEEP2 galaxies Shapley et al.,
  submitted
• KA galaxies in DEEP2 Yan et al., in prep.

First semesters data is now public
http//deep.berkeley.edu/DR1
72
Group galaxy correlation functions
We can also use correlation statistics to study
the relationship between galaxies and groups.
The group-galaxy cross-correlation shows how
galaxies are clustered within and around groups.
Red galaxies are preferentially found near the
centers of DEEP2 groups, while blue galaxies
actively avoid them. Were testing the same
thing in many ways
Coil et al. 2005, submitted, astro-ph/0507647
73
Group-based correlations are sensitive to
relationship between galaxies halos
Mock catalogs which match early DEEP2 ?(r)
predict very different clustering of group
galaxies or field galaxies than observed.
Coil et al. 2005, submitted, astro-ph/0507647
74
AGN in post-starbursts
Post-starburst galaxies with finite line emission
mostly display AGN-type line ratios. This may
provide a hint about the connection between AGN
and post-starbursts.
75
Group-based correlations provide extra
sensitivity to halo models
Our mock catalogs also predict significantly
different 1-halo and 2-halo correlations than
observed - halo radial profile appears to be off.
Coil et al. 2005, submitted, astro-ph/0507647
76
Void statistics at z1 and z0

• We have studied the Void Probability Function
  (VPF) - the probability that a sphere of radius R
  centered at a random point contains no galaxies -
  using both DEEP2 and SDSS data.

The VPF can be described by an infinite sum of
higher-order correlation functions and
potentially contains a wealth of information on
biasing, etc. However, a simple negative
binomial ansatz predicts the observed VPF very
well, given only the two-point correlation
function and the number density of the tracer
used.
Conroy et al. 2005, accepted, astro-ph/0508250
77
Measuring the VPF
Voids are larger for brighter / redder / less
common galaxies. Negative binomial ansatz
works fairly well for both DEEP2 SDSS data, for
all subsamples and scales probed.
DEEP2
Negative binomial model matches VPF for dark
matter halo centers (not mass points) in
simulations - insensitive to halo model
parameters.
SDSS
Conroy et al. 2005, accepted
78
Void statistics at z1 and z0

• We have studied the Void Probability Function
  (VPF) - the probability that a sphere of radius R
  centered at a random point contains no galaxies -
  using both DEEP2 and SDSS data.

The VPF can be described by an infinite sum of
higher-order correlation functions and
potentially contains a wealth of information on
biasing, etc. However, a simple negative
binomial ansatz predicts the observed VPF very
well, given only the two-point correlation
function and the number density of the tracer
used.
Conroy et al. 2005, submitted, astro-ph/0508250
79
Measuring the VPF
Voids are larger for brighter / redder / less
common galaxies. Negative binomial ansatz
works fairly well for both DEEP2 SDSS data, for
all subsamples and scales probed.
DEEP2
Negative binomial model matches VPF for dark
matter halo centers (not mass points) in
simulations - insensitive to halo model
parameters.
SDSS
Conroy et al. 2005, submitted
80
First DEEP2 Group Catalog
We currently have group catalogs for 3 fields
Gerke et al. 2005, astro-ph/0410721
81
Group Richness Distribution
Ngroups (sgt200 km/s)
group richness
Most groups have N2-3 within our sample (but we
are sampling L galaxies - there are many more
fainter galaxies in these groups)
Gerke et al. 2005, astro-ph/0410721
82
Why search for groups in DEEP2?
In Newman et al. (2002) we showed that the
apparent abundance of groups as a function of
redshift and velocity dispersion, dN(s,z)/dzds,
provides a useful test of the dark energy
equation of state. For modest-mass groups, this
is dominated by differences in the volume element
(which varies by 3x between w0 and w-1), though
affected by the growth factor as well.
83
Finding groups in DEEP2

• We find groups using the locations of galaxies in
  redshift space - no photometric information is
  used, just the overdensity in the 3d galaxy
  distribution.

In particular, we are using the Voronoi-Delaunay
Method of Marinoni et al. (2002), which has been
optimized for use at high z and performs well.
(For our purposes, clusters are just especially
massive groups.)
Group in early DEEP2 data ?250 km/sec
84
Changes to constraint estimates
We have recently added completely-covariant (i.e.
pessimistic) systematic errors to our constraint
estimates, and fixed an error in our growth
factor calculations noticed by Eric Linder. Here
we plot the new 95 error contours for a LCDM
model resulting from combining DEEP2 with SDSS
results, including systematic errors but assuming
s8 is known.
Tests with mocks indicate we can use groups with
sgt350 km/sec reliably.
85
Dependence on systematics
Systematic errors predominantly affect the
constraints in the w direction, which is
primarily driven by DEEP2 group abundances. SDSS
groups can still provide strong constraints in
the ?m direction (assuming s8 is known), as there
errors are small enough that the shape of the
velocity function is providing significant
constraints.
86
What if we had 20x as much area?
Future baryonic-oscillation surveys could be used
to make this same measurement if they are densely
sampled. A 60 square degree survey could yield
tight constraints on w - IF systematics are
well-constrained.
87
Constraints for w-0.7
As for most techniques, constraints are a bit
stronger for w-0.7 models than w-1.
88
We are beginning to measure w
N(s) from 314 groups are plotted. Even ignoring
redshift information, the sensitivity to w is
clear. However, the group abundance also depends
on other parameters we need to tie down
Furthermore, we are still checking systematics!!
Gerke et al. 2005
89
s8 Dependence of N(s)
The normalization of the Power Spectrum, ?8, can
strongly influence the abundance of groups, as if
?8 is greater, fluctuations are larger and groups
are more common. To be able to constrain w, we
need an accurate measurement of ?8 . New SDSS
studies are now making this possible (e.g..
Seljak et al. 2004).
90
?M Dependence of N(s)
The number of high-redshift groups is sensitive
to ?M . However, given a value of s8, the z0
SDSS cluster abundance will tie down ?M very
tightly (Newman et al. 2002).
91
Velocity bias and N(s)
A final degenerate parameter is the velocity
bias, bv. This is the factor by which the
velocity dispersion of galaxies in a cluster
differs from the dark matter dispersion. Some
simulations currently favor bv1.1, others
0.9. In the end, our results match bv1.1, ?M
0.4, ?81, or w -1.25.
92
Some conclusions

• Our results are consistent with no evolution in
  the fine structure constant from z0 to z0.7.
• Large surveys can make possible many kinds of
  scientific discoveries, and go far beyond
  whatever topics and fields are thought to be
  interesting when the survey is designed.
• We will soon have a w measurement from
  DEEP2SDSS, with 1s error in w of 0.2.
• Future baryonic oscillation surveys may be able
  to do very well at constraining evolution in ?,
  if they have the resolution and right wavelength
  coverage they will have large samples of bright,
  star-forming galaxies at z1. They could also be
  useful for group n(s,z), if we can understand the
  systematics.

93
Be wary of highest lowest zs
Shown are residuals between single-spectrum and
global fits to the perturbation to the wavelength
system from skylines. Residuals are worst at the
ends of the spectra. Quantifying now
94
Redshift Maps in 4 Fields z0.7-1.3
Cone diagram of 1/12 of the full DEEP2 sample
95
Finding groups in DEEP2

• We find groups using the locations of galaxies in
  redshift space (via VDM - cf. Marinoni et al.
  2002) - no photometric information is used, just
  the overdensity in the 3d galaxy distribution.

• Uses of DEEP2 groups include
• Cosmology N(s,z) depends on w (mostly via dV/dz)
• Galaxy formation and evolution e.g. group vs.
  field populations.

For our purposes, clusters are just the most
massive groups.
Group in early DEEP2 data ?250 km/sec
96
First DEEP2 Group Catalog
We currently have group catalogs for 3 fields
Gerke et al. 2005, astro-ph/0410721
97
Group Richness Distribution
Ngroups (sgt200 km/s)
group richness
Most groups have N2-3 within our sample (but we
are sampling L galaxies - there are many more
fainter galaxies in these groups)
Gerke et al. 2005, astro-ph/0410721
98
Measurement of w in DEEP2 Survey
N(s) from 314 groups are plotted. Even ignoring
redshift information, the sensitivity to w is
clear. However, the group abundance also depends
on other parameters we need to tie down
Furthermore, we are still checking systematics!!
99
s8 Dependence of N(s)
The normalization of the Power Spectrum, ?8, can
strongly influence the abundance of groups, as if
?8 is greater, fluctuations are larger and groups
are more common. To be able to constrain w, we
need an accurate measurement of ?8 . New SDSS
studies are now making this possible (e.g..
Seljak et al. 2004).
100
?M Dependence of N(s)
The number of high-redshift groups is sensitive
to ?M . However, given a value of s8, the z0
SDSS cluster abundance will tie down ?M very
tightly (Newman et al. 2002).
101
Velocity bias and N(s)
A final degenerate parameter is the velocity
bias, bv. This is the factor by which the
velocity dispersion of galaxies in a cluster
differs from the dark matter dispersion. Some
simulations currently favor bv1.1, others
0.9. In the end, our results match bv1.1, ?M
0.4, ?81, or w -1.25.
102
The Voronoi-Delaunay Method Group-Finder
(VDM)

• We use the Voronoi cell volume to find dense
  regions potential group seeds.
• Then we use the Delaunay mesh, its geometric
  dual, to estimate density of group core.
• Then we search adaptively for group members based
  on central density estimate.
• We have been testing VDM extensively using
  realistic DEEP2 mock catalogs to optimize the
  group-finder and test our systematics.

103
Galaxy properties in groups
We are using the group catalog to study galaxy
properties within groups. We find that redder,
early-type galaxies are preferentially found in
groups at z1, similar to local trends.
0.7ltzlt0.9
Gerke et al. 2005
104
Other recent and upcoming papers include

• Angular clustering of galaxies Coil et al.,
  2004, ApJ, 617, 765
• DEEP2 survey strategy dark energy Davis et
  al.,astro-ph/0408344
• Evolution of close-pairs/merger rates Lin et
  al., 2004, ApJ, 617, 9
• Satellite galaxy kinematics Conroy et al.,
  astro-ph/0409305
• Measuring environment in deep surveys Cooper et
  al., submitted
• Luminosity function Willmer et al. Faber et
  al., in prep.
• Group correlation function Coil et al., in prep.
• Galaxy properties vs. environment at z1 Cooper
  et al., in prep.
• KA galaxies in the DEEP2 sample Yan et al., in
  prep.
• Void statistics in the DEEP2 sample Conroy et
  al., in prep.
• Overview of the DEEP2 sample Faber et al., in
  prep.

First semesters data is now public
http//deep.berkeley.edu/DR1
105
Galaxy Properties and Environment

• We measure galaxy environments using
  projected 3rd-nearest neighbor distance, shown to
  be near-optimal in Cooper et al. 2005
  (submitted). There are strong trends of galaxy
  density with restframe color and OII equivalent
  width (a proxy for star formation rate) the
  color trend can explain the OII one.

log density
blue color red
OII equivalent width (SFR)
Cooper et al. 2005
106
Color vs. Equivalent Width of OII
Red galaxies have low OII equivalent width,
while blue galaxies span a wide range. It
appears that the scatter in this relation is most
likely not due to environment. We are currently
matching the DEEP2 colors and magnitudes to SDSS
to look for evolution in environmental dependence
between z1 and z0.
107
Galaxy Clustering in DEEP2
We are now performing a second generation of
studies of the galaxy correlation function using
volume-limited samples and a much larger dataset.
Locally, x(r) is roughly a power-law (r0/r)g
with r05 Mpc/h and g1.8. Local trends of
correlation vs. color persist at z1.
(for LgtL, z0.7-1.0, preliminary) red
r05.09 (0.11) g1.95 (0.05) blue r03.56
(0.07) g1.74 (0.05)
Coil et al. 2005, in prep.
108
x(rp,p) depends strongly on color
Red galaxies not only have a larger correlation
length, but also larger velocity
dispersion/fingers of g-d they reside in more
clustered / denser environments. We detect
coherent infall on large scales for both blue and
red galaxies.
Coil et al. 2005, in prep.
109
Clustering vs. Luminosity in DEEP2
We are starting to measure correlation statistics
of galaxies as a function of many other
properties luminosity, linewidth/velocity
dispersion, stellar mass, morphology, etc. Many
comparisons to models will soon be possible.
As at low z, brighter galaxies cluster more
strongly in DEEP2.
Coil et al. 2005, in prep.
110
Group-galaxy cross-correlation function
The group-galaxy cross-correlation shows how
galaxies are clustered within and around groups.
Red galaxies are preferentially found near the
centers of DEEP2 groups, while blue galaxies
actively avoid them. Were testing the same
thing in many ways
Coil et al. 2005a