Mapping the z 2 Large-Scale Structure with 3D Lyα Forest Tomography Intergalactic Matters Meeting, MPIA Heidelberg Max Planck Institut für Astronomie Heidelberg, Germany June 16, 2014
Collaborators: Joe Hennawi (MPIA), Casey Stark (Berkeley), Martin White (Berkeley), Xavier Prochaska (UCSC), David Schlegel (LBNL), Andreu Ariño-i-Prats (Barcelona) + Your Name Here Please see Lee, Hennawi, White, Croft & Ozbek, 2014 ApJ, 788, 49
Mapping 3D LSS with Galaxy Redshift Surveys Mapping the 3D cosmic web requires galaxy spec-z s (best photo-z s give σ z = 0.1 or δr los 100h 1 Mpc at z 2) To probe Mpc-scales require spec-z s to depths of L 0.1L. Local Universe ( v = 2000 km s 1 slice) Courtois et al 2013
Mapping 3D LSS with Galaxy Redshift Surveys Mapping the 3D cosmic web requires galaxy spec-z s (best photo-z s give σ z = 0.1 or δr los 100h 1 Mpc at z 2) To probe Mpc-scales require spec-z s to depths of L 0.1L. Local Universe ( v = 2000 km s 1 slice) z = 2.3 (COSMOS spectro-z s) Courtois et al 2013 COSMOS Collaboration
Lyman-α Forest as Probe of z > 2 Universe Lyman-α absorption caused by neutral hydrogen in front of background QSO. Each sightline probes 300 500 h 1 Mpc along line-of-sight between Lyα and Lyβ, at z 2 Q1422+2309; z = 3.63 In the optically thin IGM, the Lyα transmission F = exp( τ) traces underlying matter density, ρ dm (x)/ ρ dm, modulated by IGM astrophysics : τ(x) T 0 0.7 (x) 2 0.7(γ 1) Γ
Massive Lyα Forest Survey Data Quantity has a quality all its own, Joseph Stalin BOSS has pioneered the usage of 3D Lyα forest measurements Typical quasar sightline: g 21.5 at 15deg 2 Slosar et al 2011
Massive Lyα Forest Survey Data Quantity has a quality all its own, Joseph Stalin BOSS has pioneered the usage of 3D Lyα forest measurements Typical quasar sightline: g 21.5 at 15deg 2 Slosar et al 2011 What can be gained in going after BOSS-quality spectra on 8-10m telescopes?
Restframe UV Luminosity Functions at z 2 At g 23, the LBG luminosity function dominates over QSOs and rises steeply with observing depth.
IGM Tomography The Lyα forest in each spectrum is a 1D tracer of the IGM, but with an dense distribution of background sources it s possible to tomographically map out the IGM in full 3D (Pichon et al 2001, Caucci et al 2008). Casey Stark, UC Berkeley This will require using LBGs as background sources allowing Mpc-scale mapping. Mapping filamentary structure at 250kpc is in the science case for all the 30m-telescopes.
Source separation vs map resolution The sightline separation, d, is the basic consideration for IGM tomography. To make a map with 3D resolution ǫ 3D, we expect a requirement of d ǫ 3D. At a survey limit of g 24.5, the average transverse separation between z 2.3 sightlines is d 2.5 h 1 Mpc
Observational Requirements Moderate resolution spectrographs are sufficient, since we just need mapping scale ǫ 3D to be resolved: R > 1300 ( 1 h 1 Mpc ǫ 3D ) [ ] (1+z) 1/2 3.25
Observational Requirements Moderate resolution spectrographs are sufficient, since we just need mapping scale ǫ 3D to be resolved: R > 1300 ( 1 h 1 Mpc ǫ 3D ) [ ] (1+z) 1/2 3.25 But what is the SNR requirements for this technique? Steidel et al 2009 (Astro2010 White Paper) calls for S/N 50 per R = 5000 resolution element on R = 24.5 galaxies! In BOSS, we have been working with Lyα absorption spectra with S/N 2 per R 1500 resolution elements
Testing IGM Tomography with Sims (Lee et al 2014) Use Martin White s N-body TreePM simulation with 2048 3 particles in 250 3 h 3 Mpc 3 volume. Generate Lyα forest absorption skewers including peculiar velocities and Jeans smoothing Extract random number of sightlines corresponding to sightline density, n los Assign source magnitudes according to luminosity functions and add pixel noise assuming some t exp Right: Smoothed to R = 1000, and assuming t exp = 2hrs on Keck LRIS on g = [22.3,23.3,24.0] sources
Wiener Filtering Algorithm Wiener filtering can be applied to grid of Lyα forest skewers to reconstruct the underlying 3D field (Pichon et al 2001, Caucci et al 2008) M = C MD (C DD +N) 1 D D and M are the data and reconstructed vectors. C MD and C DD describe 2-pt correlations split into LOS and transverse parts N is noise vector this allows us to weigh data by SNR Use very fast PCG implementation written by Casey Stark (see his talk later)
Simulation of Lyα Forest Tomography at z 2.3 Tomographic Reconstruction True Lyα Forest Field Dark Matter Overdensity (100 h 1 Mpc) 2 2 h 1 Mpc slices, redshift direction is into page Smoothing scale ǫ 3D = 3.5 h 1 Mpc. Assumes survey depth of g = 24.5 and t exp = 2hrs on Keck LRIS Green dots on DM map: coeval R = 25.5 galaxies (L 0.4L ) Large rectangle: 1 sq deg; Small rectangle: Area of pilot program (see later).
3D Visualization of IGM Tomography True 3D absorption field at left; reconstruction from noisy mock spectra at right (Similar reconstruction as previous slide). Dimensions: (65 h 1 Mpc) 2 (100 h 1 Mpc)
Science with Lyα Forest Tomography Lyα forest tomography will generate LSS maps at z 2 Galaxy Protoclusters Progenitors of the largest z = 0 clusters are not necessarily the most massive halos at high-z, but at z > 2 are indeed largest overdensities on 10 15 h 1 Mpc scales (Chiang et al 2013) Should be straightforward to identify progenitor regions directly through tomographic map (Stark et al, in prep)
Science with Lyα Forest Tomography Lyα forest tomography will generate LSS maps at z 2 Galaxy Protoclusters Progenitors of the largest z = 0 clusters are not necessarily the most massive halos at high-z, but at z > 2 are indeed largest overdensities on 10 15 h 1 Mpc scales (Chiang et al 2013) Should be straightforward to identify progenitor regions directly through tomographic map (Stark et al, in prep) Clustering/Cosmology Lyα forest clustering at z > 2 has provided valuable constraints on cosmology Especially sum of neutrino masses (Σm ν ), amplitude of fluctuations σ 8, running of primordial spectral index, dn s /dk. CLAPTRAP will probe relevant 20 h 1 Mpc scales in 3D for the first time, complementary with BOSS 1D measurements.
Science with Lyα Forest Tomography Lyα forest tomography will generate LSS maps at z 2 Galaxy Protoclusters Progenitors of the largest z = 0 clusters are not necessarily the most massive halos at high-z, but at z > 2 are indeed largest overdensities on 10 15 h 1 Mpc scales (Chiang et al 2013) Should be straightforward to identify progenitor regions directly through tomographic map (Stark et al, in prep) Clustering/Cosmology Lyα forest clustering at z > 2 has provided valuable constraints on cosmology Especially sum of neutrino masses (Σm ν ), amplitude of fluctuations σ 8, running of primordial spectral index, dn s /dk. CLAPTRAP will probe relevant 20 h 1 Mpc scales in 3D for the first time, complementary with BOSS 1D measurements. Galaxy Environment Studies...
Galaxy Environments (I) Tomography will give direct constraints on DM overdensity of z 2 galaxies (modulo uncertainties in IGM astrophysics) Below right: 4 h 1 Mpc smoothed overdensity of R < 25.5 galaxies (assuming 100% completeness, perfect redshifts) Tomographic Reconstruction Dark Matter Overdensity Smoothed R < 25.5 galaxies dist.
Galaxy Environments (II) Galaxy Overdensity vs DM LyaF transmission vs DM Above: Scatter plot of volume elements in simulation box Lyα forest tomography traces DM well from 0 dm 2.5 (both smoothed to 4 h 1 Mpc) Galaxy map doesn t trace underdensities well Note: Secure R = 25.5 redshifts will need 40hr integrations on VIMOS.
Galaxy Environments (II) Galaxy Overdensity vs DM LyaF transmission vs DM Above: Scatter plot of volume elements in simulation box Lyα forest tomography traces DM well from 0 dm 2.5 (both smoothed to 4 h 1 Mpc) Galaxy map doesn t trace underdensities well Note: Secure R = 25.5 redshifts will need 40hr integrations on VIMOS. Huge opportunities to study morphology, SFR, AGN activity, color etc as function of environment at z 2
COSMOS Lyman-Alpha Mapping And Tomography Observations (CLAMATO) Survey to do Lyα forest tomography in central sq deg of COSMOS field, using Keck-LRIS and VLT-VIMOS (currently in TAC) ǫ 3D = 3.5 h 1 Mpc resolution mapping at 1000deg 2 and g 24.7 (60 h 1 Mpc) 2 300 h 1 Mpc 10 6 h 3 Mpc 3 volume Total time requirements: t exp 2hrs per LRIS pointing 160hrs total including overheads Pilot run with Keck LRIS in March 2014 through UC (PI: Schlegel) and NAOJ (PI: Lee). 2 night on COSMOS + 1 night on AEGIS lost 75% to weather
Target selection in pilot program Targeted g 24.6 (nominal) sources at 2.2 z < 3 to have Lyα forest coverage at 2.15 < z < 2.40. Spatial uniformity on the sky is emphasized. LRIS FOV is 5 7 (1/80deg 2 ) pilot area covers 0.1 sq deg 20 25 targets per mask Colors indicate forest redshift coverage NOT source redshift Symbol size reflect source mags. Largest: g 23.4; smallest: g 24.9 Square = spectroscopically confirmed source redshift ( 60%) Circle = photo-z s
Some Spectra Reduced 12 spectra from one mask with sufficient SNR for tomography. (Shapley et al 2003 LBG composite spectrum in green)
Preliminary Map Preliminary tomographic reconstruction with 12 spectra in 5 7 area of one mask, corresponding to 5 h 1 Mpc 8 h 1 Mpc transverse area. Map shows reconstructed δ F = F/ < F > 1, red is negative δ F
Comparison with COSMOS galaxies There are 7 galaxies with solid spectro-z s within the map volume Can check the δ F at the location of these galaxies. Preferentially located in higher-δ F regions: 5/7 of galaxies reside in the 30th percentile high-absorption rergions
Summary/Conclusions The first exploitation of LBGs for Lyα forest absorption analysis First 3D large scale structure map at z 1 CLAMATO: Spectroscopic survey aimed at mapping out central 1 sq deg in COSMOS with ǫ 3D 3 h 1 Mpc spatial resolution at 2.15 z 2.40 1000 LBGs + faint QSOs Aim to propose for 15 night Keck Large Program starting 2015A (through UC, 3-4 nights per semester over 2 yrs). Also proposing for VLT-VIMOS. Science: Galaxy environments, cluster progenitors, cosmology... Happy to collaborate!
Continuum Fitting Unlike QSOs, detailed physical models exist for LBG spectra (e.g. StarBurst99) LBG spectra have intrinsic absorption lines, but at moderate resolution they are not prominent in the Lyα forest region Below: best-fit Starburst99 model (solid color) fitted to MUSYC spectrum, and random Starburst99 model 5 4 Lya Object: BXmusyc_59773 z = 0 Noise f(λ) =(λ/1430a ) a ; a =-1.3 Model for the forest continium Best model: Z/Z o =2.0 Age = 4My ligther colors show other models Normalized Flux 3 2 FeII1125 FeII1142 SiII1260 OI1302 CII1334 SiIV1393 SiIV1402 SiII1526 1 0 1150 1200 1250 1300 1350 1400 1450 1500 λ [A ] Andreu Ariño (Barcelona)