Title: John Peacock Tokyo HSC Workshop Nov 2006
1Systematic challenges in future dark energy
measurements
- John Peacock Tokyo HSC Workshop
Nov 2006
2ESA-ESO Working Group on Fundamental Cosmology
John Peacock (Chair) Peter Schneider
(Co-Chair) George Efstathiou Jonathan R.
Ellis Bruno Leibundgut Simon Lilly
Yannick Mellier Major
contributors Pierre Astier, Anthony Banday,
Hans Boehringer, Anne Ealet, Martin Haehnelt,
Guenther Hasinger, Paolo Molaro, Jean-Loup Puget,
Bernard Schutz, Uros Seljak, Jean-Philippe Uzan
3Dark Energy Task Force
Report to NSF, DoE, NASA Rocky Kolb Andy Albrecht
4Cosmology Concordance Model
Heavy elements 0.03 Neutrinos 0.3 Stars
0.5 H He gas 4 Dark matter 20 Dark Energy
75
- Outstanding questions
- initial conditions (inflation?)
- nature of the dark matter
- nature of the dark energy
5The cosmological parameters
Only 6 parameters needed (flat no gravity
waves)
Normalization Optical depth from
reionization Scalar spectral index Baryon
density (? ? h2) CDM density (? ? h2) H0 /
100
WMAP3 2dFGRS
r (tensor fraction), curvature, m? , w P/?
(DE eqn of state) DE is entangled with other
parameters
6The nature of dark energy
- Zero-point energy?
- ) expect ?vac Emax4 (natural units ch1)
- But empirically Emax 2.4 meV - not a real
cutoff - (2) Dynamical Dark Energy
- Quintessence use inflationary technology of
wlt0 from scalar fields - Empirical w P/? c2 ( -1?) fit w(a) w0
wa(1-a)
7Dark Energy sensitivity degeneracy
distance
- Vacuum affects H(z)
- H2(z) H20 ?M (1z) 3 ?R (1z) 4 ?V (1z)
3 (1w) - matter
radiation vacuum - Alters D(z) via r s c dz / H
- And growth via 2H d?/dt term in growth equation
- Both effects are
- Small (need D to 0.2 for w to 1)
- Degenerate with changes in ?m
- To measure w to a few , we need to have
- independent data on ?m and to be able to
Rule of 5
growth
8Dark energy current status
Combined
Need to aim for 1 precision in future work
9Ruling in L Bayesian view
Evidence ratio trade likelihood ratio against
how much of parameter space is ruled out Trotta
(2005) roughly 1 accuracy on w will prove its
L unless we measure a large likelihood against
that model
10Evolving Dark Energy pivot redshifts
Assume w w0 wa(1-a) If observe degeneracy
w0 A Bwa, ) w A (B1-a)wa ) zpivot
1/(1B) - 1 Method zpivot CMB 0.43 BAO
z1 0.54 BAO z1z3 0.85 Lensing 0.25
Difficult to get much baseline
11Not everything may fit new physics, or
systematics?- need for multiple independent
methods - harder in future
12The CMB common basis for all methods
13CMB degeneracies and w
Comoving distance-redshift relation (? ? h2)
CMB power spectrum depends only on ?m and ?b
(apart from large-scale ISW and
reionization) Thus degeneracy between curvature
and vacuum (vary both to keep D fixed) Flat case
vary ?v and w. Degeneracy broken by large-scale
effects. Thus limit on w from good CMB data alone
14Data requirements for DM/DE
- Nearly all of the methods need redshift
information, up to and beyond z '1 - LSS/BAO need 3-D positions of 106 galaxies
spread over 3000 deg2 - Clusters quasi all-sky distribution of 105
clusters with eROSITA (104 SZ-clusters from
Planck), cannot be individually followed-up by
spectroscopy - Weak lensing needs only approximate but unbiased
redshifts of 109 galaxies over 20,000 deg2 - Supernovae have the same imaging needs, plus 8m
spectroscopy - Accurate photometric redshifts are essential,
over wide regions of the sky. - For BAO, direct spectroscopy is preferred
(photo-zs gt10 X less effective) - Near-IR is important for controlling catastrophic
photo-z failures, and is potential bottleneck
15Lensing vs SNe vs BAO (Dune)
(SNAP) (WFMOS 5000)
And similar constraints from mass function of
100,000 clusters
DETF likes ? w0 ? wa but not so clear kill ?
first
16Probe 1 Supernovae
Ground vs HST
Standard candles distances to 5 Currently
samples of few hundred Space resolution essential
for high z
17Supernovae Future challenges
- Now 500 SNe (SNLS). 2010 5,000 (Pan-STARRS)
- Could yield w to 2, but
- Photometric accuracy to lt 1 needed
- Malmquist worries at higher z
- Contamination by non-Ia
- Non-MW extinction
- Evolution of these factors, even ignoring
intrinsic evolution
18Probe 2 Gravitational lensing
Image distortion depends on - D(z) via baseline -
growth of structure
19Shear Data Ground vs Space
Typical cosmic shear is 1, and must be
measured with high accuracy Even so, 1
precision on w needs gt 10,000 deg2 surveys
Space small and stable PSF ? larger number of
resolved galaxies ? reduced systematics
20Photometric redshifts
Broad-band data can give ?z/(1z) ' 0.04 But
expect catastrophic failures for zgt1 with optical
only Sufficiently deep near-IR (K ' 22) needs
space
21Lensing future challenges
- PSF correction STEP testing shows that even
measuring shear with 1 precision is hard. - Photo-z precision need to know ltzgt in a few
tomographic z bands to 0.1 fractional accuracy - needs gt105 zs to calibrate
- Nonlinear corrections most observations to date
probe only shear correlations lt 10 arcmin - need more accurate models for nonlinear P(k)
- effect of baryons?
- Intrinsic effects
- alignment of close pairs
- foreground-background alignments
22Probe 3 Clusters of galaxies
Evolving mass function sensitive to g(z) Apparent
baryon fraction depends on D(z)
23Clusters future challenges
- Projection effects Need X-ray survey (eROSITA?)
- Mass estimates
- Accuracy of mass estimation from X-ray/SZ (8
rms) - Self-calibration to avoid bias in masses
- Redshift accuracy
- need ? z/(1z) ' 0.02
- Maybe OK via averaging photo-zs
- but needs 105 zs for calibration in any case
24Probe 4 P(k) and Baryon oscillations
- Matter-radiation horizon
- 123 (?m h2 / 0.13)-1 Mpc
- (2) Acoustic horizon at last scattering
- 147 (?m h2 / 0.13)-0.25 (?b h2 / 0.024)-0.08 Mpc
- Standard rulers to probe distance-z relation
25BAO Precision on D(z) and volume
Takada
error in D (V / 5 h-3 Gpc3)-1/2 (kmax / 0.2
h Mpc-1)-1/2
- 0.7 lt z lt 1.3 1 (h-1Gpc)3 540 deg2
- 2.5 lt z lt 3.5 1 (h-1Gpc)3 254 deg2
- Thus 1 distance accuracy (5 on w) needs 2000
(z1) or 1000 (z3) deg2 - Really prefer gt 5000 deg2 at z 1 (gt5 million
zs) z 3 selection problematic
26BAO with Photo-zs
Schuecker
Much less efficient e.g. ESO VST/KIDS, 60
million photo-z over 1500 deg2 gives 14
predicted error on w 100 times bigger sample
needed for same accuracy as WFMOS Possibly
interesting with all-sky data, but demanding on
photometric uniformity
27Systematics
- Need to show that acoustic scale is at linear
prediction to few parts in 1000 - Why might it not be?
- Mass nonlinearities
- Scale-dependent bias
- Mask convolution
- Missing close pairs
- Needs many semi-realistic mock surveys
28Semianalytic galaxies in MS 1014M halo
Dark matter
Galaxies
29Galaxy power bigger nonlinear effects at z3
than at z1
z3
z7
DM
gals
Power spectrum from MS divided by a baryon-free
LCDM spectrum
z0
z1
Springel et al. 2005
30Larger volumes mock galaxy P(k) data
Semianalytic data from Durham 1 Gpc/h cube
(ICC1000) Systematic shift in oscillation scale
relative to linear theory of 1 Can plausibly
calibrate this down to 0.1 shift
31Pan-STARRS 1
- 1.8m telescope on Haleakala, Maui
- 7 deg2 OT CCD camera
- Survey operations from mid-2007
- Initial programme runs to end 2010
- 20,000 deg2 grizy to i23.8
- 84 deg2 grizy to i27.0
- Most powerful imaging cosmology data prior to
DES, HSC - Should get w to 2-3 from each of lensing, SNe,
photo-z BAOs - Science consortium Hawaii, MPG Germany, CfA,
JHU, UK (Belfast, Durham, Edinburgh)
32(No Transcript)