Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.arcetri.astro.it/igm50/talks_all/lee.pdf Дата изменения: Wed Jun 17 11:43:38 2015 Дата индексирования: Sun Apr 10 01:28:14 2016 Кодировка: Поисковые слова: redshift survey

Mapping the z > 2 Cosmic Web with 3D Ly Forest Tomography
IGM@50 Conference, Spineto, Italy Khee-Gan ("Just call me K.-G.") Lee
Max Planck Institut fur Astronomie, Heidelberg Ё

June 8, 2015
Collaborators: Joe Hennawi (MPIA), Martin White (Berkeley), Xavier Prochaska (UCSC), Casey Stark (Berkeley), David Schlegel (LBNL), Nao Suzuki (IPMU), COSMOS collab oration
K.G. Lee Ly Forest Tomography


Mapping the Cosmic Web wth Galaxy Redshift Surveys
Ability to map LSS depends on the average galaxy separations, e.g. SDSS Main Galaxy Sample (z 0.3) has average galaxy separation 8 h-1 Mpc. At z = [0.5, 1.0, 2.0], need to go to I [22.5, 24.2, 25.7] to reach same separation.

24 deg2 VIPERS Survey on the ESO VLT, Guzzo et al 2014

K.G. Lee

Ly Forest Tomography


Mapping the Cosmic Web wth Galaxy Redshift Surveys
Ability to map LSS depends on the average galaxy separations, e.g. SDSS Main Galaxy Sample (z 0.3) has average galaxy separation 8 h-1 Mpc. At z = [0.5, 1.0, 2.0], need to go to I [22.5, 24.2, 25.7] to reach same separation.

24 deg2 VIPERS Survey on the ESO VLT, Guzzo et al 2014

Direct mapping of z > 1 LSS with galaxy redshifts only feasible with 30m te l e s c o p e s !
K.G. Lee Ly Forest Tomography


Lyman- Forest as Probe of z > 2 Universe
Restframe 1215.67 A Lyman- absorption caused by neutral hydrogen in front of background QSO. This transition redshifts into optical wavelengths at z > 2.


Credit: Michael Walther (MPIA)

We observe the transmitted flux F = f/C = exp() caused by optical depth . This absorption is seen over 300 - 500 Mpc along the quaar line-of-sight before Ly kicks in.

K.G. Lee

Ly Forest Tomography


Ly Forest as a Probe of the Cosmic Web
In the modern 'fluctuating Gunn-Peterson' model, the Ly absorption traces the quasi-linear matter overdensity, dm (x)/ dm , probing the range 0 10. This is modulated by IGM astrophysics
-0.7

(x)

T

0



(x)

2-0.7(-1)

IGM temp erature at mean

density, T

0

UV background ionization rate, Temp erature-density relationship,

(where T () -1 )
Credit: AmSci/R. Simcoe

K.G. Lee

Ly Forest Tomography


Ly Forest as a Probe of the Cosmic Web
In the modern 'fluctuating Gunn-Peterson' model, the Ly absorption traces the quasi-linear matter overdensity, dm (x)/ dm , probing the range 0 10. This is modulated by IGM astrophysics
-0.7

(x)

T

0



(x)

2-0.7(-1)

IGM temp erature at mean

density, T

0

UV background ionization rate, Temp erature-density relationship,

(where T () -1 )
Credit: AmSci/R. Simcoe

In this talk, I assume that the Ly forest traces large-scale structure not `gas' !
K.G. Lee Ly Forest Tomography


Ly Forest Tomography
Collection of closely-separated sightlines enable tomographic reconstruction of 3D absorption field on scales comparable to sightline separation (Pichon et al 2001, Caucci et al 2008, Lee et al 2014)

Credit: Casey Stark (Berkeley)

K.G. Lee

Ly Forest Tomography


Source separation vs map resolution
The sightline separation, d , is the basic consideration for IGM tomography. For maps with 3D resolution 3D , expect to need d 3D .

K.G. Lee

Ly Forest Tomography


LBGs as Ly Forest Background Sources?
QSO luminosity function (Palanque-Delabrouille +2013) rises too slowly to provide sufficient background sources to sample the Ly forest.

With LBGs, exp onential increase of source density, n separation is d 2.5 h-1 Mp c at g 24.5.
K.G. Lee

los

10m . Source

Ly Forest Tomography


High-Resolution Spectrum of a Lensed LBG
The z = 2.724 LBG MS 1512-cB58 is lensed ( 50в) to V 20.6

Savaglio et al 2001 (12hrs exposure per order on VLT-UVES)
K.G. Lee Ly Forest Tomography


Observational Requirements for Ly Forest Tomography

LBGs are much fainter than QSOs (m 24), so it was assumed that 30m-class telescopes would be required for tomography, but never any detail analysis of requirements. In Lee et al 2014a (ApJ 788, 49), I argued using sims and analytic c a l c u l a ti o n s :
No need to resolve individual Ly absorb ers: R 1000 is adequate S/N 3 - 4 p er A is sufficient at the survey limit At reconstruction scales of 3D > 2 h-1 Mp c, still in shot-noise limited


regime so d < 3D is helpful

K.G. Lee

Ly Forest Tomography


Simulated Tomographic Reconstructions
Reconstructions using mock sp ectra with realistic sampling and sp ectral S/N. Transverse maps smoothed on = 3.5 h-1 Mp c scale (left to right): n n n
los los los

= 971deg = 657deg = 112deg

-2 -2 -2

, te , te , te

xp xp xp

= 8hrs = 6hrs = 2hrs

(Exp osure times assuming VLT VIMOS sp ectrograph)


100Mpc /h



K.G. Lee

Ly Forest Tomography


Pilot Observations on Keck, March 2014
Observing run with LRIS spectrograph on 10m Keck-I telescope, Hawai'i. Suffered 70% weather loss, but from 4hrs on-sky obtained 24 LBG spectra at 2.3 < z < 2.8

Brightest: g = 23.84

Median: g = 24.39

Faintest: g = 24.74

K.G. Lee

Ly Forest Tomography


COSMOS/CANDELS/3D-HST Field

6 M p c /h
x 6 5
perp



4

(z=2.325) (h-1Mpc) 3 2

1

0 14

22

[2.462, 24.39] [2.457, 23.98]

[2.465, 24.39]

13
[2.554, 24.73]



12
[2.551, 24.52]

20
[2.615, 24.26] [2.864, 24.60] [2.553, 24.28]

11

10

[2.503, 24.63]

9
[2.810, 24.86]

18

8
[2.408, 24.25] [2.505, 24.75] [2.720, 24.28] [2.917, 24.07]

7 (z=2.325) (h-1Mpc) y

14 Mpc/h

Declination

16
[2.438, 24.53]

6

5
[2.646, 24.69] [2.431, 24.43]

14
[2.435, 24.17] [2.706, 24.35] [2.321, 23.84] [2.305, 24.26]

4

3

2

12
[2.685, 24.71]

1

perp

0 10
[2.826, 24.00]



-1

-2
[2.690, 24.89]

02° 08 10h 00m 45
s

-3 20
s

35s 30s 25s Right Ascension Center: R.A. 10 00 32.41 Dec +02 15 17.9 40

s

K.G. Lee

Ly Forest Tomography


Tomographic Reconstruction
We now have extracted transmission F = F/ F - 1 ('data'), pixel noise estimates F , and [x, y, z] positions. Perform Wiener filtering on these i n p u ts to e s ti m a te th e m a p : M=C
MD

· (C

DD

+ N)-1 · D

The noise term provides some noise-weighting to the data. We assume Gaussian correlation function in the map, where CDD = CMD = C(r1 , r2 ), and C(r1 , r2 ) = 2 exp - F with L = 3.5h
-1

(r )2 (r )2 exp - , 2 2L 2L2
-1

(1)

Mpc and L = 2.7 h

Mpc, and F = 0.8.

K.G. Lee

Ly Forest Tomography


Tomographic Reconstruction
We now have extracted transmission F = F/ F - 1 ('data'), pixel noise estimates F , and [x, y, z] positions. Perform Wiener filtering on these i n p u ts to e s ti m a te th e m a p : M=C
MD

· (C

DD

+ N)-1 · D

The noise term provides some noise-weighting to the data. We assume Gaussian correlation function in the map, where CDD = CMD = C(r1 , r2 ), and C(r1 , r2 ) = 2 exp - F with L = 3.5h
-1

(r )2 (r )2 exp - , 2 2L 2L2
-1

(1)

Mpc and L = 2.7 h

Mpc, and F = 0.8.

Super-efficient implementation by Casey Stark (Berkeley), see (arXiv:1412.1507), I just wait 1 min on my laptop....

K.G. Lee

Ly Forest Tomography


First 3D Map of Cosmic Web at z > 2

V = (6 в 14) h-2 Mp c2 в 230 h-1 Mp c 1.93 в 104 h-3 Mp c3 (27 h-1 Mp c)3

K.G. Lee

Ly Forest Tomography


Pilot Map in Slices
0 < xperp (h-1 Mpc) < 2
yperp (h-1 Mpc)

3950
12 9 6 3 0

4000

Comoving Distance (h-1 Mpc) 4050

4100

4150

2.25

2.30 z 2 < xperp (h-1 Mpc) < 4

2.35

2.40

2.45



rec F

0.40 Comoving Distance (h-1 Mpc) 4050 0.15 4100 4150 -0.10 -0.35 -0.60 4100 4150

yperp (h-1 Mpc)

3950
12 9 6 3 0

4000

2.25

2.30 z 4 < xperp (h-1 Mpc) < 6

2.35

2.40

2.45

yperp (h-1 Mpc)

3950
12 9 6 3 0

4000

Comoving Distance (h-1 Mpc) 4050

2.25

2.30 z

2.35

2.40

2.45

Squares: 18 coeval galaxies (mostly zCOSMOS-Deep) with known spectro-z's within map, error bars are estimated 1 redshift errors.

K.G. Lee

Ly Forest Tomography


Pilot Map in Slices
1 0 -1

0 < xperp (h-1 Mpc) < 2 3950 4000 Comoving Distance (h-1 Mpc) 4050 4100 4150

yperp (h-1 Mpc)

12 9 6 3 0

F

2.25
1 F 0 -1

2.30 z 2 < xperp (h-1 Mpc) < 4

2.35

2.40

2.45



rec F

0.40 Comoving Distance (h-1 Mpc) 4050 0.15 4100 4150 -0.10 -0.35 -0.60 4100 4150

yperp (h-1 Mpc)

3950
12 9 6 3 0

4000

2.25
1 F 0 -1

2.30 z 4 < xperp (h-1 Mpc) < 6

2.35

2.40

2.45

yperp (h-1 Mpc)

3950
12 9 6 3 0

4000

Comoving Distance (h-1 Mpc) 4050

2.25

2.30 z

2.35

2.40

2.45

Overdensities seen in the map are typically probed by multiple i n d e p e n d e n t s i g h tl i n e s

K.G. Lee

Ly Forest Tomography


Pilot Map in Slices
0 < xperp (h-1 Mpc) < 2
yperp (h-1 Mpc)

3950
12 9 6 3 0

4000

Comoving Distance (h-1 Mpc) 4050

4100

4150

2.25

2.30 z 2 < xperp (h Mpc) < 4
-1

2.35

2.40

2.45



rec F

0.40 Comoving Distance (h-1 Mpc) 4050 0.15 4100 4150 -0.10 -0.35 -0.60 4100 4150

yperp (h-1 Mpc)

3950
12 9 6 3 0

4000

2.25

2.30 z 4 < xperp (h-1 Mpc) < 6

2.35

2.40

2.45

yperp (h-1 Mpc)

3950
12 9 6 3 0

4000

Comoving Distance (h-1 Mpc) 4050

2.25

2.30 z

2.35

2.40





Mp 14

c /h

2.45

25 Mpc /h

Hints of a huge overdensity at z=2.43?

K.G. Lee

Ly Forest Tomography


A Large Protocluster at z = 2.44?

Credit: Yi-Kuan Chiang (UT Austin)

Color Scale Overdensity of photo-z candidates Stars HETDEX Pilot Survey (Chiang et al, submitted) Squares zCOSMOS spectro-z's from Diener et al 2015

K.G. Lee

Ly Forest Tomography


A Large Protocluster at z = 2.44?

Credit: Yi-Kuan Chiang (UT Austin)

Color Scale Overdensity of photo-z candidates Stars HETDEX Pilot Survey (Chiang et al, submitted) Squares zCOSMOS spectro-z's from Diener et al 2015

See Yi-Kuan Chiang's p oster!
K.G. Lee Ly Forest Tomography


Do Galaxies Live in Overdensities on 3 - 4 Mpc Scales?
We evaluate the F values at the 18 galaxy p ositions, and compare with the full map. Bias towards higher overdensities, but also a few in lower-density regions.
(a) Data Reconstruction
0.25 PDF of Reconstructed Map 0.20 0.15 0.10 0.05 0.00 -0.8
Full Map Galaxies

-0.6



-0.4

-0.2



rec F

0.0

0.2

0.4

0.6

Higher densities

This is due to redshift uncertainties in the foreground galaxies + reconstruction errors. More volume + galaxies needed for better measurements!

K.G. Lee

Ly Forest Tomography


Comparison with Simulations
We took Ly forest skewers from sims, and created mock data with exactly the same sightline geometry and S/N as real data. We also have p ositions of DM halos corresp onding to R 25.5 (abundance-matched)
Mock Spectra
4040 14 12 (h-1 Mpc) 10 8 6 4 2 0 2.32 2.34 z 2.36 2.38 Comoving Distance (h-1 Mpc) 4060 4080 4100



rec F

0.40 0.15 -0.10 -0.35 -0.60

y

perp

Idealized Spectra
4040 14 12 (h-1 Mpc) 10 8 6 4 2 0 2.32 2.34 z 2.36 2.38 Comoving Distance (h-1 Mpc) 4060 4080 4100



rec F

0.40 0.15 -0.10 -0.35 -0.60

y

perp

Simulation
4040 14 12 (h-1 Mpc) 10 8 6 4 2 0 2.32 2.34 z 2.36 2.38 Comoving Distance (h-1 Mpc) 4060 4080 4100



3D F

0.60 0.20 -0.20 -0.60 -1.00

There are reconstruction errors, but our data quality should reproduce

broad LSS features
Redshift errors will also scatter galaxies out of overdensities
K.G. Lee Ly Forest Tomography

y

perp


CLAMATO Survey
(COSMOS Lyman-Alpha Mapping And Tomography Observations)
Prop osed survey targeting 0.8 sq deg of COSMOS field ( 30 nights on

Keck)
Target 1000 LBGs at 2.3 Simulation Field

z 3 for R 1000 sp ectroscopy z 2.3 LSS map over 106 h-3 Mp c3 (100 h-1 Mp c)3
Mo ck CLAMATO Reconstruction

Dimensions: (65 Mp c)2 в (100 Mp c)
K.G. Lee Ly Forest Tomography


K.G. Lee

Ly Forest Tomography


CLAMATO: State of the Union
Keck/LRIS Run 2015 April 16-21: 4 masks targeting z = 2.15 - 2.40 and 1 mask targeting z = 2.09 protocluster
2.40

Ly forest coverage: z=2.15 (red) & z=2.40 (blue)
Lee+14 Map z=2.45 PC (Diener+14)

2.35

2.30 Dec (deg)

2.25

2.20

2.15 2.10 150.2

z=2.09 PC (Yuan+14)
150.1 RA (deg) 150.0 149.9

Upcoming large program proposal beginning 2016A.
K.G. Lee Ly Forest Tomography


Hunting Protoclusters with Ly Tomography
Stark, White, Lee & Hennawi (arXiv:1412.1507): studied progenitors of simulated M > 1014 M clusters at z = 2.5
z = 0 DM z = 2.5 DM z = 2.5 Ly

M = 9 в 1014 M

M = 3 в 1014 M

Protoclusters are r 5 Mp c overdensities in Ly absorption CLAMATO will find M > 3 в 1014 M progenitors with 90% purity and

75% completeness N 5 in 106 h-3 Mp c3

Even with known protoclusters, can characterize full 3D morphology, e.g.

collapsing along single axis vs more isotropically
K.G. Lee Ly Forest Tomography


Detecting High-z Voids with CLAMATO
Stark, Font-Ribera, White & Lee (submitted): look for LSS voids with simulated Ly forest tomography
DM z-space DM `True' Flux Reconstructed Flux

R = 11.7 h-1 Mpc void, Stark+2015

Used simple spherical finder: grow spheres around minima until some . Ї CLAMATO will b e able to pick up R

6 h-1 Mp c voids with 70% purity and 60% 100 voids within CLAMATO volume

Synergy with JWST-NIRSPEC to study sub-L void galaxies in the

z 2 - 3 accretion era?
K.G. Lee Ly Forest Tomography


Voids and Protoclusters in CLAMATO
Right: Central part of COSMOS Field
Magenta CLAMATO 0.8 sq deg Orange CANDELS/3D-HST footprint
Dec (deg) 2.4 -40 2.6



65 h

-1

Mp c


40 40

Comoving Separation at z=2.3 (h-1 Mpc) -20 0 20

CANDELS/3D-HST


20 Comoving Separation at z=2.3 (h-1 Mpc)

Lee+14

65 h
-1

Blue Lee+2014 Pilot field Dots Photo-z and spectro-z z = 2.4 - 3.0 LBG targets

2.2

0

Mp c

2.0

-20 1.8

Below: Simulated protoclusters and voids (approx to scale)
Protocluster


-40 150.6 150.4 150.2 RA (deg) 150.0 149.8

Void

K.G. Lee

Ly Forest Tomography


Other Science with CLAMATO


Finding and characterizing high-z protoclusters Finding and characterizing high-z voids Galaxy Environments: How do high-z galaxy properties (e.g. SFR, metallicity, AGN activity) correlate with their large-scale IGM environment Decomposing high-z LSS into filaments, sheets and nodes First measurement of LSS topology at z > 2, e.g. genus or Euler characteristics Cross-correlating Ly forest with LBGs and LAEs: first detections of RSD from these population + better bias measurements Small-scale 3D power spectrum of the Ly forest: independent constraints on cosmological parameters e.g. 8 and m Refining photometric redshifts with tomographic LSS as a prior
K.G. Lee Ly Forest Tomography










Pushing Tomography Towards Smaller Scales
8-10m class telescop es can p erform IGM tomography down to scales of 3 - 4 h-1 Mp c or 1.5 - 2pMp c, too coarse to resolve IGM/CGM interface ( 100pkp c). This will require 30m-class telescop es.

K.G. Lee

Ly Forest Tomography


Pushing Tomography Towards Smaller Scales
8-10m class telescop es can p erform IGM tomography down to scales of 3 - 4 h-1 Mp c or 1.5 - 2pMp c, too coarse to resolve IGM/CGM interface ( 100pkp c). This will require 30m-class telescop es.
Right: Exposure time required to achieve different tomographic resolutions with GMT/GMACS Different curves are different `map SNR' (e.g. below) defined by Var(true)/Var(map-true). 4hr integrations will push down to 500pkpc scales
`True' SNR=3.0

Exposure Time (hrs) Resolved tomographic scale (Mpc/h)
SNR=2.5 Ly Forest Tomography

K.G. Lee


Protocluster Galaxies as Background Sources
In addition to studying protoclusters in the foreground, we can also use known protocluster galaxies as background sources for high-resolution tomography in the foreground.
COSMOS Field HS1549+19 z = 3.1

Average source separation of g < 25 galaxies is d 1.5 h protocluster sources vs d 3 - 4 h-1 Mpc in field.
K.G. Lee Ly Forest Tomography

-1

Mpc from


Detailed IGM Mapping through Protocluster Lines-of-Sight (DIMPLS)
Reconstructions resolving scales of 400kp c, directly detect cosmic web filaments at z 2.6
Tomographic Map from Mock Data
6 5 4 3 2 1 0 0
rec F
0.60

Transverse Distance (h-1 Mpc)

0.20

10

20 Line-of-Sight Distance (h-1 Mpc)

30
-0.20

Simulation `Truth'
6 5 4 3 2 1 0 0
-0.60

-1.00

10

20 Line-of-Sight Distance (h-1 Mpc)

30

Simulated reconstructions resolving 3D 1.5 h-1 Mp c (400pkp c) Up coming observations:
35hrs (Priority `B') on HS1549 with VLT-FORS2 6 nights on SSA22 with Keck-DEIMOS
K.G. Lee Ly Forest Tomography


Summary


First exploitation of LBGs as Ly forest background sources


Faintest-ever (g 24.5 vs g 21.5 for BOSS) and densest-ever source densities ( 1000 deg-2 vs 10 deg-2 for BOSS)





First large-scale structure map of the z > 2 universe (from 1/2 n i g h t o f d a ta ! ) Ongoing CLAMATO survey:






Survey 1000 LBGs (z 2 - 3) in 0.8 deg2 field Will yield 3D Ly forest tomographic map with 3 h-1 Mp c spatial resolution over (100 h-1 Mp c)3 Time requirement: 30 nights on Keck-LRIS (inc weather overhead) Science: cosmic web studies at z > 2, hunting galaxy protoclusters and high-z voids, galaxy prop erties as function of environment....



Over the next few years, use protoclusters as background sources to probe IGM-CGM interface before 30m-class telescope come online All simulation products available at http://tinyurl.com/lya- tomography- sim- data
K.G. Lee Ly Forest Tomography


Continuum Estimation
The Ly forest transmission F = f/C is observed flux, f, divided by estimated intrinsic `continuum', C. Fortunately there are few strong absorbers in the Ly forest region, which we can mask.

Berry et al 2012

We perform `mean-flux regulation' (Lee et al 2012) using the Berry et al 2012 composite at 1040 - 1190 A, i.e. adjust amplitude and slope until the resulting F matches measurements from quasars.
K.G. Lee Ly Forest Tomography



UV spectra of Low-z Starforming Galaxies

Rivera-Thorsen+ 2015
K.G. Lee Ly Forest Tomography