The PAU (BAO) Survey

Enrique FernandezUAB / IFAE Barcelona

43rd Recontres de Moriond (La Thuille, March08)

The PAU (Physics of the Accelerating Universe) Project

•Large (~8,000 dg2, 0.1<z<0.9) photometric galaxy survey with purposely-built camera.

•Project has been proposed by 7 Spanish institution to a special program of the MEC (ministry of science). The team (40 persons) includes astrophysicists, cosmologist and particle physicists (experimenters and theorists).

•Funding (5 years) for the camera and other activities.

The PAU (Physics of the Accelerating Universe) Project

•Large (~8,000 dg2, 0.1<z<0.9) photometric galaxy survey with purposely-built camera.

•Project has been proposed by 7 Spanish institution to a special program of the MEC (ministry of science). The team (40 persons) includes astrophysicists, cosmologist and particle physicists (experimenters and theorists).

•Funding (5 years) for the camera and other activities.

The PAU Project

•We still need to settle on

- A telescope with a large fraction of the observing time (the goal is to complete the survey in 4 years). Several options being considered.

- More collaborators (camera, survey itself, ...).

The PAU Project

•Focus on measuring the Baryon Acoustic Oscillations peak, in both angular and radial directions.

•Simulations show that we can obtain a precision on z for LRG (luminous-red galaxies) of

z ~ 0.003 (1+z)

•There will be a wealth of other physics that can be studied with the survey data.

7

• Galaxy redshift surveys are used to measure the 3D clustering structure of matter: for BAO only need position and z, no flux, no shape.

• There can be several sources of systematic errors:

– Light from galaxies is a “biased” estimator of matter content

– Non-linear physics involved in galaxy formation

– Redshift distortions

• However, all effects tend to predominantly change the amplitude of the correlations, but not the position of the measured acoustic peak

BAO from Galaxy Redshift Surveys

• BAO are quite insensitive to systematic errors. In any case, the systematic errors are very different from those of SNe.

• But the effect is small only visible at large scales which leads to huge surveys.

BAO measured in SDSS data (Eisenstein et al. 2005)

h = H0 / (100 km s-1 Mpc-1) ~ 0.7

3.5- detection of BAO at <z> = 0.35(confirmed by 2DF and SDSS photometric surveys at about 2.5 )

)(

)(,)()()( 21

rrrrr

dP12 2 1(r) dV1dV2,r r 1

r 2

or

Based on 55000 “luminous red galaxies” from the SDSS spectroscopic galaxy survey

Dark energy and BAO

z

A zH

dz

z

czddz

zH

czdr

0 )'(

'

1)(

)()(

BAO gives us a standard distance with a co-moving value

rBAO~ 100 Mpc/h (rBAO= 146.8±1.8 Mpc, CDM)

For a flat universe

radial angular

)(

)( zd

r

zH

rz

A

BAOBAO

BAOBAO

Importance of measuring in the radial direction:

)1(3

320

2

)1)(1(

)1()(

wm

m

z

zH

zH

Assume flat universe, w=constant and m=.25

Error propagation:

11

Cosmological Results from BAO

SDSS BAO: Eisenstein et al. 2005

SNLS SNe: Astier et al. 2006

12

)(

11

1

)(

)(

knPVkP

kP P(k): power spectrum

n: galaxy density

• Statistical errors on galaxy-galaxy correlation functions are determined by “sample variance” and “shot noise”.

• Sample variance: how many independent samples of the relevant scale (150 Mpc)3 one has volume

• Shot noise (Poisson): how many galaxies included in each sample density

Feldman, Kaiser, Peacock, ApJ 426,23 (1994)

Size and resolution requirements

The required Volume and the required precision in z were studied with two detailed N-body dark matter simulations done by the MICE collaboration using the GADGET-2 code : (Fosalba, Castander, Gaztañaga,Manera, Miralda-Escudé, Baugh, Springel; http://www.ice.cat/mice)

Lbox Npart halo mass acronym Mpc/h number 1011Msun/h Nhalos

MICE3072 3072 20483>375 1.1x106

MICE1536 1536 10243 >47 2.1x106

LCDM model with m=.25, =.25, b=.044, ns=.95, 8=.8, h=.7; ns=2.4x1011Msun/h; L=50Kpc/h

14

Size and resolution requirements

)(

11

1

)(

)(

knPVkP

kP

For the scales of interest for PAU (LRGs, 0.1<z<0.9) nP(k)>10, so that the Poisson term is negligible.

It can also be shown that:

VGpchV

rBAOBAO /8%6.05.0 33

2/13

We aim at 1% error in BAO V=8h-3Gpc3 Area =8,000 deg2

We expect about 14M LRG, with L>L* above IAB=22.5 in the sample.

15

To study the required precision in z the two-point correlation function of over 1M halos with M>3.7x1013h-1Msun was studied.

The position of the halo was smeared with a Gaussian:

)(

)1(z

2

1-exp ~)( z

2

zH

cz

z

rrf z

z

Size and resolution requirements

Real space, perfect resolution

z-space, perfect z-resolution + peculiar velocities

z-space, z = 0.003(1+z) + peculiar velocities

z-space, z = 0.03(1+z) + peculiar velocities

Visual illustration of the importance of z resolution

linear corr. func. (b=3)

non-linear (RPT; Crocce-Scocimarro, 2008)

z = 0.003 (1+z)

z = 0.007 (1+z)

x z = 0.03 (1+z)Curves are analytical predictions derived from

P(kt,kz)=PNL exp [-kz2 z

2]

Fosalba

Requirements on Redshift Precision

/)( zd A

zzH /)(/1

z / (1+z) z / (1+z)

H(z)

dA(z)

spec specphoto photo

Inve

rse

of a

rea

of w

0-w

a er

ror

ellip

sePadmanabhan

The PAU SurveyPhotometric survey. Target “Luminous Red Galaxies” as in many other surveys. These are old elliptical galaxies, which are very bright and have a characteristic spectrum with a prominent break at 4000Å.

The position of the peak gives us z.

20Benítez

The PAU Survey: use a filter system consisting of ~40 filters (100Å wide), plus two wide filters (similar to SDSS u and z)

survey at Calar Alto

20 filters

Moles et al.

Expected z resolution

From back-of-the-envelope calculation (assume step-function in flux, falling between two filters):

05.04000

2

FFff

z

for =100Å filters

z = 0.003 (1+z) at z=0.9 f /F=0.12 S/N~12,

which is achievable for LRG at this redshift.

Expected z resolution

Much more elaborated simulation:

- Exposure time calculator with observing conditions taken from several sites

- CCDs as in DES (LBL CCDs)

- Filters as in Alhambra

- 2m telescope; 6 deg2 FoV camera

- optimization of exposure times

- Galaxies brighter than IAB=23

- Model for LRG Bruzual&Charlot (11Gyr, Z=0.2)

- Photo-z’s from BPZ (Benítez)

Use odds parameter from BPZ photo-z method to eliminate badly determined z’s.

z ph

ot-z

s/(1

+z s

)

In red the LRGs for which the odds is less than 0.5.

The r.m.s. of the remaining LRGs is well within the 0.003(1+z) limit

Benitez

z r.m.s as a function of the true z

Benitez

Spatial density of LRG with IAB<23

n(z) > 10-3 (h/Mpc)3

Telescope-camera system: 2m-class telescope with a ~ 6 deg2 FoV camera ~ 500 Mpixels with 0.40”/pixel ~60 CCDs 2Kx4K.

This is demanding but feasible.

Alternative is to place camera in an existing (larger diameter) telescope of smaller aperture. Possibility of using dichroic mirrors also being explored.

PAU instrument

Conceptual design studies for a telescope with the required parameters exist (from industry), as well as cost estimates.

28

Comparison with Other BAO Surveys

Padmanabhan

29

Dark Energy Parameters

Miquel

Miquel

Conclusions

For the measurement of BAO a resolution in z of the of

(z) = 0.003 (1+z)

is close to optimal.

A survey of 8,000 deg2, from 0.1<z<0.9 will give ~ 14 M LRG. From this sample the BAO scale can be measured both in the angular and radial (z) directions to 1%. This results in a substantial improvement of standard cosmological parameters, making it a competitive survey with respect to those being planned at present.

This precision can be obtained photometrically with a multi-filter system of about ~40 filters, 100Å wide.

Back up

32