Ludovic Van Waerbeke

1
Progress in Weak Lensing by Large Scale
Structures from CFHTLS to SNAP
Ludovic Van Waerbeke Department of Physics and
Astronomy University of British Columbia
2
The theoretical paradigm L-CDM Wm 0.25, Wb
0.04, WL 0.7, h 0.7, plus random phase np
0.95 fluctuation spectrum

• Great successes
• Fairly accurate cosmological parameter
  measurements
• Self-consistent estimates for P(k) over 4.5 dex
  in scale from various experiments.

SNLS, Astier et al. astro-ph/0510447
WMAPall sets, Spergel et al. astro-ph/0603449
3
High precision cosmology from WMAPALL (Spergel
et al. 2006, astro-ph/0603449)
High precision Cosmology Are we done?
4
What next? Prospect for the next decade.
Is Cold Dark Matter the correct picture?
Hierarchical model, galaxy formation (feedback,
cooling and merging), dark matter halo
distribution. Problems with small scale power,
light halo size and their rotation, merging
history (very massive galaxies are in place at
high redshift) Is Dark Matter truly non
interacting? What is the equation of state of
Dark Energy? Is inflation the correct picture?
i.e. predicting DM power spectrum. Is G.R. the
correct theory of gravitation at large scale?
5
Cosmological weak lensing
1deg.
Wide field imaging survey
6
DARK MATTER
Lensing could be used -to calibrate the mass
of Halos for various halo sizes -to establish
catalogues of Mass selected clusters -to study
in an unbiased way the Mass/ligh relation
WHAT HAVE WE LEARNED SO FAR?
7
EDisCS -- the ESO Distant Cluster Survey --
White et a. 2004, Clowe et al. , 2005
20 VLT/FORS2 fields preselected to contain
distant galaxy clusters between redshift 0.4 to
1. Photometric, spectroscopic and morphology
studies.
8
(No Transcript)
9
Open symbols low Versus high-z X-ray Selected
clusters. (Clowe et al. 2006)
13 () out of 20 clusters have optical
counterpart and show evidence of
velocity Dispersion dependence with
redshift. The 7 remaining candidates have no
obvious optical counterpart (s gt 3.5)
10
Athreya et al. 2001 dark matter study of MS1008
Dark matter map shows a bimodal mass
distribution, which the X-ray emission Is located
between the two peaks.
Many examples of mass/shape/substructure
measurements
11
1E0657-56 The bullet cluster (Clowe et al.
astro-ph/0608407 )
Dark Matter is made of non-interacting
dissipationless objects.
12
Projection effect (large scale structures
environmental effects) White, LVW, McKay, 2002
Bias and offset in the measured Mass of 1014 Mo
clusters.
Weinberg Kamionkowski (2002) have shown that
20 of the 2-3 s peaks Are dark because they
correspond to unvirialised structures and/or
projection. Cen Ostriker (1999) and Dave
(2001) proposed the existence of large cool gas
of clouds, the WHIM with no stars that could be
as massive as a cluster of gal.
13
Best dark clump candidate (Erben et al. 2001)
A1942
ROSAT
HST
Chandra
Lensing
Counts CFHT
Counts HST
14
The shape of dark matter haloes
Concentration parameter cvirial radius/scale
radius4 (Cold Dark Matter)
A1689 (Oguri et al. 2005), CL0024 Kneib et al
(2003), and MS2137 (Gavazzi 2005).
Concentration parameter MUCH higher than
Predicted by CDM C14 instead of a few. A1689
is triaxial
A new crisis in CDM Paradigm?
Mass bias? Selection Bias? Biased mass Function?
15
Dark Matter clustering history is an important
cosmological tool
Today
z2
Bartelmann et al. 2003
16
Statistical lensing
17
Searching for the excess of variance against the
random alignment of distant galaxies, which is
attributed to gravitational lensing.
18
Cosmic Shear two-points Statistics and mass Power
Spectrum
Convergence (or projected mass) power spectrum
Top-hat shear variance at scale Qc
Aperture mass (Map) variance at scale Qc
Shear correlation function at separation Q
19
The gravitational potential is a scalar field,
shear must NOT be rotational
E projection Pure gravity signal
45 degrees rotated galaxies, E -gt B
20
What do we probe with lensing?
(To first order, strong degeneracy with power
spectrum normalisation and shear calibration)
21
Cosmic shear surveys and dark energy P(k) what
angular scale do we need to probe ?

• Primordial power spectrum P(k) ? kn
• Normalisation ?8
• Shape evolves with time cosmic history of
  structure formation and cosmological parameters

Caldwell, Dave, Steinhardt 1998 Ma, Caldwell,
Bode, Wang 1999
Cosmic variance??

• Important scales
• 50 lt l lt 105
• Beyond 10 degrees
• less usefull or w
• cosmic variance
• Optimal
• 50 lt l lt 103
• range not sensititive to the non-linear
  evolution models of dark matter power spectrum
• Goal 10 lt l lt 5o
• Small scales data for halos formation history

• Non-linear evolution
• Peacock Dodds??
• Smith et al ??

22
(WMAPCBI) Cosmic Shear CFHTLS Orthogonal
degeneracies

• Cosmic shear CMB
• Spectral index and running spectral index test
  of inflation models.
• Model CFHTLS MCMC
• 170 deg2,
• n20 gal/arcmin2
• ??0.4,
• zs0.8
• scales 0.6 to 2 deg.
• marginalised over the 5 other parameters

Tereno, Doré, van Waerbeke, Mellier 2004
23
Statistical shear 2004 compilation
24
VIRMOS Z0.9 LVW et al.
RCS Z0.6 Hoekstra Et al.
25
The Canada France Hawaii Telescope Legacy Survey
26
E and B modes on the WIDE data
Aperture mass and top-hat variance
Correlation function
27
Comb ined constraints with WMAP3 (Spergel et al.)
and CFHTLS WIDEDEEP (Benjamin et al. in prep.)
Non-gaussian covariance Semboloni et al. 2006.
28
DIRECT TECHNICAL ISSUES (data analysis)
29
Recent results with the Canada France Hawaii
Telescope Legacy survey
DEEP data Semboloni et al. 2005 WIDE data
Hoekstra et al. 2005
Megacam
30
Shape measurement methods impressive diversity
Unpublished, simulated cosmic shear survey
analysed with KSB
Before correction
After correction
Good recovery of the PSF anisotropy, Problem with
the isotropic correction? This is something that
will not show up in the E/B analysis.
Survey simul No shear
Survey simul With shear
31
Image simulations provide crucial tests of galaxy
shape measurement
Kuijken 2006 is the most Recent analysis and
shows A precision of 1-2 percents Is achievable
with current Analysis techniques.
32
From Kuijken 2006 shows the size and magnitude
dependence of the Residual bias.
33
(No Transcript)
34
Photometric redshift
Spectro and photoz Compilation
CFHTLS-DEEP Ilbert et al. 2006
35
Source redshift distributions use external
data Set if photometric redshift not available
Wide data
Deep data
ltzgt1
ltzgt0.8
N(z) calibrated from Hubble Deep Fields
36
Fernandez-soto et al. 2000
Deep data, using u,g,r,I,z photoz
High and low redshift bins. First tentative of
tomography as proposed by Jain Taylor,
independent of structure growth
37
The Deep fields are observed in u,g,r,I,z,
but still this is not good enough for Photoz!
Bolzonella et al. 2000
38
Generic calibration requirement as function of
stastical noise and cosmic variance (LVW et al.
2006, astro-ph/0603696)
No degradation in signal implies a mean source
redshift accuracy of 0.5 In redshift bins
(Huterer et al. 2006).
39
Calibration requirement for selected lensing
experiements
40
HIGHER LEVEL ISSUES (data interpretation)
41
Estimation of the Non-linear power spectrum
Cold Dark Matter simulations Smith et al. 2004
Is there really a universal fitting formula?
42
Do we have to include the baryons?
Two competing effects will change the mass power
spectrum Baryon cooling and heating
White 2004, baryon cooling
Knox et al 2004 baryon heating
43
Intrinsic alignment issue
No doubt there is intrinsic galactic alignment in
3D space?
Pen, Jounghun, Seljak (2000), Spin-Spin
correlation 12000 spirals at cz lt 5000km/s
Brown et al. (2002), SuperCOSMOS, 10000sq.deg.,
bJlt20.5, z0.1
44
Intrinsic alignment issue
Croft Metzler (2000) galaxies are DM halo
N-body simulations, signal 1/20 lensing
for zS1, q few arcmin
Heavens et al. (2000) galaxies are thin disks
perpendicular to DM halo
angular momentum. N-body
simulations, signal 1/10 lensing for zS1, q lt
10
Catelan et al. (2001) 1) galaxy ellipticity
follows halo shape (tidal field).
2) ellipticity follows angular
momentum (quadratic in F)
leads to stronger small scale
signal and generates
B mode.
Crittenden et al. (2001) extend 2) in Catelan et
al. and make a full analytical Treatment of the
problem. Signal 1/100 lensing for z1 and q lt
10
Jing (2002) 5123 N body simulation, signal
lensing for zS1, q few arcmin
Two orders of magnitude discrepancy between
different predictions No evidence for detection
so far
45
Intrinsic alignment issue
Van den Bosh et al. (2003) significant mixing
between gas and DM halos (SPH GADGET , Springel
et al.)
46
Shear-Intrinsic correlation (Hirata Seljak
2004, Mandelbaum et al. 2005, Heymans et al.
2006)
Measured on SDSS, Shows that elliptical
galaxies Are preferentially aligned with Their
hosting DM haloes.
47
Impact on cosmic shear contamination
Could be controled using Redshift informatio
(King 2005)
NEED PHOTOzs. how well You can separate lensing
from Intrinsic and lensing-intrinsic Signals is
still not known.
48
Source Clustering impact still largely unknown.
Requires photometric redshift to be corrected.
Exemple of contamination on the skewness of the
convergence (5-40 effect)
49
PROSPECTS
50
Cross correlating mass and light (Van Waerbeke
1998)
Hoekstra et al. 2002

• Ratio

• Galaxy-galaxy, galaxy-mass, mass-mass correlations

• Indication for stochasticity at
• small angular scale?

51
3D dark matter inverstion on the Virmos-Descart
data (Pen et al. 2003)
3D inversion from 2D spectra (galaxies DM),
using broad bands estimates (Map)
Tegmark Zaldarriaga (2003)
52
High order statistics skewness
S3ltK3gt/ltK2gt2 40 W-0.8 zs-1.3
Convergence histogram
Simulated 3x3 degrees convergence Map (Jain et
al. 2000)
LVW et al. 1999
W-L constraints
53
High order statistic of the mass distribution
probe of the gravitational instability
3
Bernardeau et al. 2003, Pen et al. 2003
2
1
Evidence for a low density universe Independent
on the mass power spectrum
Wmlt0.5 at 90
3-pts function pattern
3
2
d12
1
Simulated 3x3 degrees convergence map
54
Complete shear 3-pts correlation function
analysis on simulated data
Takada Jain 2005
55
Takada Jain 2004 Shows the gain on
parameter Estimates adding high order statistics
Dark Energy constraints With lensing and PLANCK
56
Running spectral index with weak lensing and
PLANCK
57
Statistical lensing on maps, peak stat, high
order, etc
Mass reconstruction and peak statistic
Miyazaki et al., 2002
Subaru 2.1 sq.deg. field
Peak PDF
Gaussian field
Peak statistics (LVW 2002)
58
Breaking degeneracies from 25 deg2 to 1000 deg2
Future
Present
From Hu 1999

• CFHTLS (MegacamWircam) intermediate 200 deg2
• 1000 deg2, is the minimum goal for post-CFHTLS..
  A challenge
• It must be done in both visible and NIR Another
  challenge!!
• It must probe higher scale and have better S/N.
  Go to space!!!

59
On going/future cosmic shear surveys
CFHTLS 5 bands ugriz COSMOS 2sq.deg, 2 bands ig
numerous follow ups Deep Lens Survey
7x4sq.deg. CTIO, 5 bands Subaru Large-aperture
Synoptic Survey Telescope 8.4m, 7 sq.deg. Fov,
(1000sq.deg. I24)
2010 PANSTARR 4x1.5m, 2007-2010 VST 2.5m
Omegacam 1sq.deg (shallow 1000sq.deg.)
paranal VISTA 4m paranal, optical
imager? SUBARU ? Wide field optical
imager Dune/JDEM/SNAP ?
In terms of image quality, we still dont know
how good space I, but we know it is always
better than the ground (work in progress)
60
The big question space versus ground DETF
report
61
Lensing from space

• SNAP, a wide field imager in space
• Sloan Digital Sky Survey coverage with Hubble
  Space Telescope image quality

62
The largest optical survey from space COSMOS
treasury survey (2 deg2 in two filters with ACS
on HST)
63
Comparison of SNAP vs COSMOS
SNAP
Cosmos single filter
areas are to scale!
64
What should you get away from this
talk? Precision cosmology is still a dream but
close to reality. Lensing are essential this is
the only way to see the dark matter at low
redshift You can draw thousands of Fisher matrix
contours to make predictions, but There is
difficult work that remains to be done data
analysis, data interpretation We know EXACTLY
what to do, but working power is lacking!