The next decade of weak lensing science Rachel Mandelbaum, CMU.

The next decade of weak lensing science

Rachel Mandelbaum, CMU

R. Mandelbaum 2

CosmologyA homogeneous and isotropic

universe

Spatially flat and expanding (accelerating!)

General Relativity:Function of the metric (defining space-time behavior)

Stress-energy tensor

describes matter/energy

contents

R. Mandelbaum 3Picture credits: NASA/WMAP science team

Name for model: CDM

?????

??

4Picture credits: NASA/WMAP science team

Quantum fluctuations seed small (/ ~10-5) inhomogeneities…

…which are imprinted in CMB...

Matter domination: growth through gravitational instabilityR. Mandelbaum

5

Two classes of cosmological probes

Geometric: SN1A, BAO

Growth of structure

Picture credits: ESA/ESO (left), MPE/V. Springel (right)

R. Mandelbaum

R. Mandelbaum 6

Summary: current status in cosmology

An observationally supported big picture

BUT… many fundamental uncertaintiesnature of DM and DE, nature of inflationary era, GR confirmation on many scales. ?

R. Mandelbaum 7

A key problem:The universe is dominated by dark

contents.

But…we cannot directly observe those contents using a telescope.

8

Gravitational lensing

Lensing deflection of light:

R. Mandelbaum 9

Sensitive to all matter along line of sight, including dark matter!

R. Mandelbaum 10

Weak lensing

Unlensed

Lensed

Galaxies aren’t really round

NASA, ESA, S. Beckwith (STScI) and the HUDF Team

R. Mandelbaum 13

Cosmic shearShape autocorrelation statistical map of large-scale structure

R. Mandelbaum 14

Galaxy-galaxy lensingStacked lens galaxy position – source galaxy

shape cross-correlation

Reveals total average matter distribution around lens galaxies or cluster (galaxy-mass correlation)

R. Mandelbaum 15

State of the field of weak lensing

R. Mandelbaum 16

Subaru telescope

8.2 meter primary mirror

Mauna Kea

Excellent imaging conditions

R. Mandelbaum 17

Subaru telescopeMany instruments for optical and spectroscopic

observations, e.g. Suprime-Cam

Miyatake, Takada, RM, et al (2012)

18Picture credit: S. MiyazakiR. Mandelbaum

R. Mandelbaum 19

HSC is on the telescope!

HSC blog at naoj.org

R. Mandelbaum 20

Looking good!

R. Mandelbaum 21

3-layer HSC survey

Wide: ~1400 deg2, i<25.8 (grizy) Weak lensing, z<1.5 galaxy populations

Deep: ~26 deg2, 1 mag deeper, 5 wide+3 NB filters Ly-α emitters, quasars, deeper galaxy

populations, lensing systematics, …

Ultradeep: 3 deg2, 1 mag deeper, 5 wide+6 NB filters Supernovae, galaxies to z<7

Important synergies: CMB (ACT+ACTPol), redshifts (BOSS + assorted other), NIR, u band, …

R. Mandelbaum 22

What has driven this development?

~8-10 years ago, people started to realize how very powerful cosmic shear is as a probe of dark energy!

(LSST science book)

R. Mandelbaum 23

What has driven this development?

~8-10 years ago, people started to realize how very powerful cosmic shear is as a probe of dark energy!

Zhan et al. (2006, 2008)

R. Mandelbaum 24

A reminderCosmic shear measures the matter power

spectrum

This is easily predicted from theory (modulo small-scale effects)

Contrast: the galaxy power spectrum from redshift surveys – galaxies are a biased tracer of matter

Position

GalaxiesDensity

Dark matter halo

R. Mandelbaum 25

BUT

R. Mandelbaum 26

This is actually kind of difficult.

Cosmic shear is an auto-correlation of shapes:

Multiplicative biases are an issue!Coherent additive biases become an

additional term!

R. Mandelbaum 27

That’s not the only problem, either.

Intrinsic alignments

Theoretical uncertainties on small scales (e.g. baryonic effects)

Photometric redshift uncertainties

R. Mandelbaum 28

ImplicationsAs datasets grow, our control of systematics

must get increasingly better

The past ~3 years have seen a change of perspective within the lensing community: We should measure cosmic shear But we should also identify combinations of

lensing measurements with other measurements that allow us to calibrate out / marginalize over systematics directly

Use ALL the information available Minimize the combination of statistical +

systematic error!

R. Mandelbaum 29

What data will we have?The lensing shear field: HSC

The 2d galaxy density field: HSC

(Sometimes) 3d galaxy density field and velocity field, with spectroscopy: BOSS

X-ray (galaxy clusters): XMM

SZ (galaxy clusters), CMB

lensing: ACT

Lensing magnification field?

(M. White)

R. Mandelbaum 30

Summary of approach to future

data: Cross-correlate everything with

everything= more information= less sensitivity to observational uncertainties specific to one particular method

R. Mandelbaum 31

What about galaxy-galaxy lensing?

Typically undervalued for cosmology, because it measures gm correlations, not mm

Observationally easier: Coherent additive shear errors do not contribute

at all! (cross-correlation) Intrinsic alignments:

Don’t enter at all, with robust lens-source separation

If sources are not well behind lenses, they contribute, but in a different way from cosmic shear

R. Mandelbaum 32

Observational quantities

• ξgg from galaxy clustering

• ρξgm from g-g weak lensing

• Infer matter clustering (schematically): Constrain

nonlinear matter power spectrum on large scales

R. Mandelbaum 33

Let’s include cosmic shear

Use cosmic shear (mm), galaxy-galaxy lensing (gm), and galaxy clustering (gg)

Dependence on intrinsic alignments, shear systematics: Different for the two lensing measurements

Joachimi & Bridle 2011, Kirk et al. (2011), Laszlo et al. (2011) showed that the cosmological power is = that of cosmic shear, even when marginalizing over extensive models for systematics!

R. Mandelbaum 34

A concrete example:

Lensing + clustering in SDSS DR7

(RM, Anze Slosar, Tobias Baldauf, Uros Seljak,

Christopher Hirata, Reiko Nakajima, Reinabelle Reyes,

2012)

R. Mandelbaum 35

Observational quantities

• ξgg from galaxy clustering

• ρξgm from g-g weak lensing

• Infer matter clustering (schematically): Constrain

nonlinear matter power spectrum

Cross-correlation coefficient between galaxies, matter

R. Mandelbaum 36

Problem: small scalesTheoretical uncertainties in Σ (surface

density): Baryonic effects Cross-correlation ≠ 1 Cannot remove by avoiding small scale ΔΣ

Integration lower limit is the problem

R. Mandelbaum 37

Solution to small-scale issues

• Define “Annular differential surface density” (ADSD):

NO dependence on signal below R0!

→0 at R0

→ΔΣ at R>>R0

T. Baldauf, R. E. Smith, U. Seljak, RM, 2010, Phys. Rev. D, 81, 3531RM, U. Seljak, T. Baldauf, R. E. Smith, 2010, MNRAS, 405, 2078

R. Mandelbaum 38

Example from simulationsC

ross

-corr

ela

tion

coeff

(r c

c)

Using ΔΣ

Using ϒ, R0=3 Mpc/h

Reconstruction

ϒmm

R. Mandelbaum 39

Sensitivity to cosmology

Fiducialcosmolog

y:

Ωm=0.25σ8=0.8ns=1.0

R. Mandelbaum 40

Results

Lenses: SDSS-I spectroscopic samples: LRGs, z~0.3, typically 3L*, ~105

Main, z~0.1, typically L*, 6 × 105

Sources: 6 × 107 fainter galaxies

Treat samples separately, for sanity checks

Updated treatment of lensing systematics (RM et al. 2011, Reyes et al. 2011)

R. Mandelbaum 41

Example of current data

Stacked data:

~105 LRGs (lenses),

70M sources

Lensing

signal

Transverse separation R (Mpc/h)

R. Mandelbaum 42

Lensing data

R. Mandelbaum 43

Clustering data

R. Mandelbaum 44

Actual procedure

Direct fitting: Nonlinear power

spectrum PT-motivated

parametrization of non-linear bias

With these data alone, fitting for σ8, Ωm, and bias, marginalizing over bias and lensing calibration: σ8 (Ωm/0.25)0.57 =

0.80±0.05

R. Mandelbaum 45

Non-flat, free wde

R. Mandelbaum 46

Comparison to cosmic shear results

COSMOS (Schrabback et al. 2010), 11% σ8 constraint

CFHTLenS (Kilbinger et al. 2012), 4% σ8 constraint

Typical z~1, 0.8 vs. 0.25 for SDSS

SDSS gives better control of redshift systematics

Results shown here establish SDSS among the most

competitive extant surveys for weak lensing cosmology!

R. Mandelbaum 47

Near future improvements

BOSS + HSC:Less dominated

by lensing statistical errors

R. Mandelbaum 48

But that’s not all…Small-scale lensing profiles reveal galaxy DM

halos

Transverse separation R (Mpc/h)

R. Mandelbaum 49

Example of how we can use this: FoG

• Small-scale effect due to velocity dispersion within halos• Cannot simply

eliminate by using only individual halos, unless chosen “center” is really at center

White et al. (2011): contours of 3d correlation

function

R. Mandelbaum 50

Idea for how to calibrate out FoG

Hikage, Takada, Spergel (2011)

Rely on spectroscopic / photometric survey synergy

Select halos, then compare several measurements for different choices of halo centers: Redshift-space power spectra Galaxy-galaxy lensing (matter distribution) Photometric galaxy cross-correlation

R. Mandelbaum 51

ModelingNeed HOD model for how galaxies populate

halos

Include variable fraction that are offset within halos, their spatial and velocity distributions

Hikage, RM, Takada, Spergel (2012)

R. Mandelbaum 52

The punch line

40% (70%) of bright (faint) LRGs are actually off-centered satellites

Typical off-centering radius of 400 kpc/h

Typical velocity dispersion: 500 km/s

R. Mandelbaum 53

ConclusionsCurrent g-g lensing measurements:

Test theory predictions for galaxy-DM relationship

Constrain cosmological parameters at various redshifts

Lensing is the ONLY technique that directly probes the total matter distribution!

Future datasets: better S/N cosmologically interesting powerful constraints on growth of structure, done optimally via combination of multiple observables