spectroscopic surveys beyond desi · redshift surveys increasing 10x every 10 years 140 billion...

D. Schlegel, DESI@FNAL 27 May 2015 1

SpectroscopicSurveysBeyondDESIDavidSchlegel,BerkeleyLab

Outline

2

- Redshi=surveyobjec?ves:mappinglinearmodes- DESIgoals+technologies- BeyondDESI

- DESI-II:200Mgalaxiesatz<2- FOBOS:Lyman-alphaatz>2

- Investmentstokeepus“onthecurve”

Redshift survey objectives: Map all the linear modes

3

Cosmologicalinforma?oncontentisinthelinearregime

non-linear modes

linear perturbations on scales > 10 Mpc at z=0

Redshift survey objectives: Map all the linear modes

4

Font-Ribera, McDonald, Slosar in prep.

Cosmologicalinforma?on(darkenergy,primordialPk,neutrinos)containedintheperturba?vemodesofthemaps

How many modes are there?

5

CMB ~few million

modes

z=1066

~2billionlinearmodesfrom0<z<4(over20,000sqdeg)1.9millionmodestobemeasuredbyDESI2.1millionmodestobemeasuredbyLSST

6

z=1066

CMB ~few million

modes

~2billionlinearmodesfrom0<z<4(over20,000sqdeg)1.9millionmodestobemeasuredbyDESI2.1millionmodestobemeasuredbyLSST

How many modes are there?

samefigure,logscale

7

How many galaxies to measure these modes?

10milliongalaxies0<z<0.4120milliongalaxies0<z<1.52billiongalaxies0<z<4

Redshift surveys necessary to map all modes

8

Redshi=surveysΔcz~300km/satz=0,preservesinforma?ondowntonon-linearregime(e.g.,smallscales)

Redshift surveys necessary to map all modes

9

PhotometricsurveysΔcz~10,000km/satz=0,washesoutmanyofthelinearmodes;notrecoverableevenw/numbers

Low-resolution / photo-z redshifts pay a penalty

10

Spa?alresolu?onhelpsto1”SpectralR>3000greatlyreducesskylevelbyremovingskylines

(relevantforground-basedinstruments)SpectralR>3000improvesredshi=s/n

Redshift surveys are necessary to map all modes

11

Therearemoreobjectsofinterestonaspectroscopicfocalplanethanonanimagingfocalplane~10,000/deg2forimaging→informa?onhassaturated~100,000/deg2forspectroscopy

12

HST Ultra-Deep Field 10,000 galaxies / (11 arcmin2)

Redshift surveys increasing 10X every 10 years

140 billion

1980 20611000

Year

log N(galaxies)SDSS, 2009 929,000

CfA1, 1983 1840

2dF, 2003 221,414

CfA-2, 1998 18,000

LCRS, 1996 18,678

SDSS-III, 2014 2.8 million

DESI 30 million

All linear modes mapped by ~2043 All detectable galaxies mapped by ~2061

4meterprimary1meterdiamcorrector5000fiber-robotarmy200,000metersfiberoptics10spectrographsx3cameras

DESI Goals + Technologies

14

DESI Goals

DESI

4 million LRGs

17 million ELGs

0.7 million Ly-A QSOs +1.7 million QSOs

10 million BGS galaxies

35milliongalaxy+QSOredshi=survey

15

How many galaxies to measure these modes?

10milliongalaxies0<z<0.4120milliongalaxies0<z<1.52billiongalaxies0<z<4

—>DESIwillmap~100%ofthese—>DESIwillmap~10%ofthese

integral

16

DESI Technologies

6-lensop?calcorrector,1-mdiameter,includesADC

17

DESI Technologies

Focalplatemoun?ng5000fiberrobots

TherobotarmyofDESIreplaceshand-pluggingoffibers

19

DESI Technologies

10spectrographsX3cameras/spectrograph

20

DESI Technologies

Forward-modelingofspectrodataofferssubstan?alimprovementsovertheold-school,20th-century,datareduc?onofSDSS-I

SDSS-IoperatedatS/N~hugeSDSS-III/BOSSoperatedatS/N~50DESIwilloperateatS/N~10

“data”

log 1

0 [ p

ixva

l / <

pixv

al>]

Model fiber PSF for

SDSS1 @ 8500Å

The future of redshift surveys?

Modest improvements in capability from SDSS-I → SDSS-III/BOSS → DESI

22

Howhascapabilityimproved?• CCDs→improved,esp.inthered/infrared• CCDelectronics→lowernoise• Stabilityofcalibra?onsystems

•allowingbeoersky-subtrac?on,fainterobjects

Biggergainshavebeeninmul?plexing• Op?caldesignsforwiderfieldsontelescope&inspectrographs• Morefibers,hand-plugged→massively-parallelrobots

Technical challenges are cost effectiveness

23

Pre-SDSS→SDSSrevolu?onMul?plexingfrom1-30objects→640objects

usingfiber-fedspectrographsIn2015,SDSShascollectedmoregalaxyredshi=s(2.7M)

thanallothertelescopesonearthcombined

Cost models for multi-objects spectrographs scaling from DESI

24

Telescope+corrector($40M+$10M)*(Mirrorarea/4-m)2

Focalplane+spectrographs$1000XNfiber+$1MX(Nfiber/500)

Opera?ons$6M/yearX5years

+$200k+$50k

+$150k

+$300k

Costfor1%improvement

DESIcostmodelisreasonablywell-balancedMostimprovementwouldcomefrommorefibers,

butnotpossibleonDESIgivenotherdesignconstraints

25

140 billion

1980 20611000

Year

log N(galaxies)SDSS, 2009 929,000

CfA1, 1983 1840

2dF, 2003 221,414

CfA-2, 1998 18,000

LCRS, 1996 18,678

SDSS-III, 2014 2.8 million

DESI 30 million

Future investments to keep us “on the curve”?

Future investments to keep us “on the curve”?

26

Fiberrobotcosts→Pick-and-playfiberposi?oners(2dF,MMT,…)donotscale→Fiberrobotsw/1robotperfiberscales

~$10,000/fiberforSubaru/FMOSin~2006~$2000/fiberforSubaru/PFSin~2013~$700/fiberforDESIin~2015

Futureinvestments?→DOEstartedR&Din2006withLBNLLDRDsupport,followedbyDESIdevelopment→Con?nuedR&Dwouldfurtherreduceper-robotcosts

Two possible near-term redshift survey concepts

28

Op?on1:DESI@KioPeak,LSST@CerroPachon10,000sqdegfootprintequatorial

Op?on2:DESI@CerroTololo14,000sqdegfootprintequatorial+south

DESI-II + LSST for z<2

29

DESI-II + LSST for z<2

30

1. Retain DESI @ Kitt Peak with 10,000 sq deg overlap, or move DESI instrument from Kitt Peak -> CTIO • Optics study from Tim Miller (16 Sep 2014) shows no

changes needed to the optics (in fact, it’s better there!) 2. Re-furbish instrument with faster electronics

• Probably necessary for shorter exposures 3. Target ~200,000,000 galaxies with the best photo-z’s

• LSST will have a parent sample of ~10 billion galaxies 4. Turn photo-z’s ➔ spectroscopic redshifts

• Δz~0.03 ➔ Δz~0.001

DESI-II + LSST for z<2

z

χr2

trialzz2nd best

χr2

∆too∆ z

small

parabola fit toestimate error δz

best

The trick: DESI operates at S/N > 7 DESI-II at S/N > 3

Bolton, Schlegel et al. 2012

FOBOS @Keck for z>2

31

FOBOS @Keck for z>2

32

1. Utilize Cass or Nasmyth focus at either Keck telescope 2. 1000-fiber robotic focal plane, using DESI technology 3. DESI spectrographs 4. High-density Lyman-alpha forest mapping at z > 2

Map from CALMAT program K.G. Lee

Longer-term redshift survey concept

What’s possible with current tech?

34

1. Optical design from LAMOST/Guoshoujing telescope • ~2X aperture of DESI

2. DESI fiber positioners • 5X fiber positioners (25,000)

3. Survey speed increase 10X for galaxies at z < 2, 5X for Lyman-alpha forest at z > 2

Summary

35

Linearmodesforcosmology—DESIwillmap~100%ofmodesatz<0.4

~10%ofmodesatz<1.5~2%ofmodesatz<4

—DESI-II+LSSTcouldmap100%ofmodesatz<1.5—“BeyondDESI”needonlymap~2billiongalaxiesatz<4

Ifthephotonsarevaluable…Spectroscopyiscost-effec?veuseofthosephotons

ForthisCosmicVisionsProcess…DESI-II+LSSTredshi=ingshouldbedemonstrated/simulated